# Agentlist — complete Markdown edition

> An open, source-linked directory of 52 persistent personal AI agents — systems that keep context about one person and act on their behalf. Every comparison field cites a dated primary source, and a field no source answers is recorded as `unknown` rather than as a "no". Edition 1, snapshot 2026-10-26, sources checked 2026-10-21 to 2026-10-26. Descriptions are read from official documentation, not independently tested. Code and original catalogue contributions are MIT-licensed; product marks belong to their owners and linked source text keeps its original rights — credit "Agentlist" with a link when reusing this.

52 agents (14 Self-hosted runtime, 30 Hosted agent, 2 Coding tool (adjacent), 2 Under review, 4 Local-first app), 33 comparison fields in 6 groups, 51 dated updates, 3 recorded ecosystem relationships, 19 personal data sources. Disclosure: Project creator affiliated with Ambi; uniform evidence rules and an editorial featured default order.

This file contains the whole catalogue: the method, every field definition, the directory table and all 52 agent records. Nothing needs to be fetched to read it. Per-agent files live at https://agentlist.wiki/agents/<id>.md; the structured dataset is at https://agentlist.wiki/data.json.

---

# How Agentlist works

> Scope, evidence rules and the contribution process behind 52 personal-agent records. Edition 1; sources checked 2026-10-21 to 2026-10-26.

## What counts as a personal agent

A personal agent maintains context about one person over time and uses tools to act on their behalf. Three properties decide inclusion: it remembers (preferences, people, projects or a durable workspace survive one task), it can act (tools, apps, a browser or a computer, under a person's direction), and it can return (schedules, events or continuing goals bring it back between conversations).

Hosted products, local apps and self-hosted runtimes are all in scope. A new version of a product updates the existing record rather than adding one; aliases keep the old name searchable. A separate record exists only for a distinct product. Coding harnesses are catalogued as adjacent and hidden by default: the distinction is their product scope, not whether they have any persistence.

- **Catalogue types:** `managed` Hosted agent; `runtime` Self-hosted runtime; `local` Local-first app; `watchlist` Under review; `adjacent` Coding tool (adjacent)

## Evidence rules

Every source carries an ID, a title, a direct URL and the date it was checked; every claim references those IDs. Product facts come from official product pages, documentation, repositories, app listings or announcements, paraphrased rather than quoted. Where sources disagree, the product's own official page wins over third-party coverage, which goes stale as products reprice and rename — except for facts a vendor has reason to shade (funding, certifications, independent feedback), where a primary filing or the certifying body is the better source.

A claim nobody has established is omitted from the record and filled in by the build as `unknown`. Descriptions are not independent performance tests.

## What documented, limited and unknown mean

- `documented` — Supported by official documentation; not independently tested.
- `limited` — Only partial or conditional support established.
- `unknown` — Not established by reviewed sources; does not mean unsupported.

Capabilities that look similar are kept apart: a public API, an MCP connector, browser automation, a password vault and a saved browser session are five different answers. Computer control is three independent questions, and a cloud computer says nothing about whether the agent can touch your desktop.

## Dates

Source dates are the day the source was checked, per agent, so re-checking one product never restates the freshness of the rest. The catalogue reports the span of those dates rather than a single date. Launch events accept year, month or day precision (`2026`, `2026-02`, `2026-02-03`) and a day is never invented. The earliest non-update event is the launch date. A stated future ship date is recorded as an expected date on the availability object, never as a launch.

- **Launch kinds:** `launch` Public launch; `preview` Research preview; `beta` Public beta; `announcement` Announcement; `release` Open-source release; `update` Major update

## Companies and funding

The organization names the team that makes the product; the company object records its type and stage separately from product availability. Types are editorial groupings, not rankings — open-source software can still come from a funded startup. Funding is a dated, disclosed round at company level: round amount is kept apart from total raised and from valuation, and committed or pre-acquisition capital is labelled. Stage and funding are absent when not established, and each carries its own evidence.

- **Company types:** `public` Public company; `lab` Major AI lab; `startup` Startup; `established` Established company; `independent` Independent project; `research` Research group

## The comparison fields

33 fields in 6 groups.

### Company & launch

Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.

- **Company type** (`companyType`)
- **Company stage** (`companyStage`)
- **Disclosed funding** (`funding`)
- **Launch date** (`launchDate`)

### What it can do

What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.

- **What stands out** (`uniqueFeatures`)
- **Computer use** (`computerUse`)
- **Controls your computer** (`onYourComputer`)
- **Its own computer** (`agentComputer`)
- **Works while your computer is off** (`whileYouAreOff`)
- **Browser control** (`browserUse`)
- **Voice & ambient listening** (`voice`)
- **Background work & triggers** (`scheduling`)

### Apps & access

Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.

- **Native desktop apps** (`desktopApps`)
- **iPhone & Android apps** (`mobileApps`)
- **Web & messaging** (`channels`)
- **Messaging services** (`messaging`)
- **Native hardware** (`hardware`)
- **Developer API** (`api`)

### Models & plans

What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.

- **Models & providers** (`models`)
- **Bring your own model** (`byom`)
- **Included in another plan** (`bundledPlan`)
- **Price & usage costs** (`pricing`)

### Vendor lock-in

How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.

- **If the vendor stops** (`worksWithoutVendor`)
- **Remote shut-off** (`remoteDisable`)
- **What you keep** (`dataAfterExit`)

### Ownership & control

What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.

- **Open source & license** (`license`)
- **Where it runs** (`hosting`)
- **Passwords & credentials** (`passwordManager`)
- **Approval before actions** (`approvals`)
- **What it remembers** (`memory`)
- **Edit & export your data** (`portability`)
- **Privacy & isolation** (`security`)
- **Availability & access** (`availability`)

## Structured vocabularies

- **Availability states:** `available` Available; `waitlist` Waitlist; `reservation` Reservation; `announced` Announced; `unverified` Access unverified
- **Availability restrictions:** `plan` Requires a paid plan; `region` Limited regions; `age` Age restricted; `rollout` Gradual rollout; `invite` Invite required
- **Computer control · Controls your computer:** `full` Controls your computer; `partial` Limited access to your computer; `none` No access to your computer
- **Computer control · Its own computer:** `cloud` Has its own cloud computer; `local` Runs on hardware you own; `none` No computer of its own
- **Computer control · Works while your computer is off:** `yes` Keeps working; `no` Stops; `partial` Partly continues
- **Watching the agent's own screen:** `takeover` Watch and take over; `view` Watch only; `none` No remote access
- **Vendor lock-in · If the vendor stops:** `yes` Keeps working; `partial` Partly keeps working; `no` Stops working
- **Vendor lock-in · Remote shut-off:** `none` Cannot be switched off remotely; `possible` Can be switched off remotely; `required` Requires a licence or activation check
- **Vendor lock-in · What you keep:** `self-serve` Export it yourself; `on-request` Export on request; `partial` Partial export; `none` No export described
- **Hardware states:** `available` Shipping hardware; `optional` Optional hardware; `announced` Announced, not shipping; `none` No hardware; `unknown` Not established
- **App and messaging statuses:** `available` Available; `beta` Beta; `preview` Preview; `announced` Coming soon; `limited` Limited availability
- **Native platforms:** `watchos` Apple Watch; `web` Web; `macos` macOS; `windows` Windows; `linux` Linux; `ios` iOS; `ipados` iPadOS; `android` Android; `mobile` Mobile app
- **Messaging services:** `imessage` iMessage; `rcs` RCS; `sms` SMS; `whatsapp` WhatsApp; `telegram` Telegram; `signal` Signal; `slack` Slack; `discord` Discord; `teams` Microsoft Teams; `wechat` WeChat; `wecom` WeCom; `feishu` Feishu; `matrix` Matrix; `email` Email
- **How open a data source is to third parties:** `open` Open to third parties; `gated` Open, with review; `portability` EU portability only; `closed` First-party only; `none` Reachable by nobody
- **How an agent reaches a data source:** `first-party` First-party; `official-api` Official API; `portability` EU portability; `automation` Browser automation; `none` Cannot reach it
- **Ecosystem relationships:** `built-on` Built on; `fork` Code fork; `renamed` Renamed
- **Update types:** `release` Release; `launch` Launch; `research` Research; `availability` Availability change

A messaging integration never establishes a native app, and a listed product is never a shipping product: announced hardware, a paid reservation and an open download are three different answers.

## Discussion and feedback

Discussion entries link original dated threads with a short summary, and separate company or creator posts from independent community discussion. The sentiment on a thread describes that thread, not every user of the product.

Aggregate feedback snapshots are separately attributed, with provider, collection window, source channels, rated user-week counts and the raw positive share. The displayed percentage is that raw share, not the provider's composite score, and an area breakdown names the provider build it came from because those numbers change with each build. Snapshots from different dates are never mixed.

## Updates and the ecosystem graph

Dated updates and relationships are curated: collection gathers candidates into an ignored review inbox and nothing publishes automatically. The ecosystem graph carries only runtime dependencies, code forks and renames; inspiration, shared model providers, similar names and competing alternatives stay out of it, and fork ancestry needs explicit evidence.

**Open research queue:** Copilot Autopilot (private preview, tenant-scoped); OpenClaw rename history; Verified Git fork ancestry; Permissions and memory export tests.

## Contributing

- One JSON file per agent in `data/agents/<id>.json`; the filename matches the stable, lowercase, hyphenated ID.
- Run `npm run add:agent -- example-agent "Example Agent" https://example.com`, or copy `templates/agent.json`. The generator never overwrites an existing agent.
- Fill in the summary, company context, launch dates, claims, plans, platforms, messaging and sources, each claim referencing its source IDs.
- Save an official PNG, SVG or WebP mark in `src/logos/` and record its original URL and the official page. Logos belong to their owners.
- Run `npm run verify` — build, data validation and the domain tests — and open a pull request. Maintainers review before merge; CI runs the same checks.

Repository: https://github.com/sunnya97/agentlist. Generated `dist/` files, including this one, are not committed.

## Independence and licence

Disclosure: Project creator affiliated with Ambi; uniform evidence rules and an editorial featured default order.

Code and original catalogue contributions are MIT-licensed. Product marks remain the property of their owners, and third-party source text, linked posts and feedback data keep their original rights and attribution. Credit this catalogue as "Agentlist" with a link when you reuse it.

Entry point for machines: https://agentlist.wiki/llms.txt. Everything in one file: https://agentlist.wiki/llms-full.txt. Structured dataset: https://agentlist.wiki/data.json.

---

# Agentlist — the directory

> Every agent in the catalogue: 52 records, sources checked 2026-10-21 to 2026-10-26. `unknown` means no reviewed source answered that question — never that the answer is no, and `(limited)` marks an answer only partly established.

| Agent | Maker | Availability | Price | Where it runs | What you keep | Record |
| --- | --- | --- | --- | --- | --- | --- |
| Agent Zero | Agent Zero, s.r.o. | Available | Free; you pay models | Self-hosted | Partial (limited) | [agent-zero.md](https://agentlist.wiki/agents/agent-zero.md) |
| Alexa+ | Amazon | Available · some regions | Free with Prime; US$19.99/mo | Provider cloud | On request | [alexa-plus.md](https://agentlist.wiki/agents/alexa-plus.md) |
| Ambi | Ambi | Available | Trial, then paid | Unverified (limited) | On request | [ambi.md](https://agentlist.wiki/agents/ambi.md) |
| Bond | BOND App Inc. | Available | Free; US$49/mo | Provider cloud | Self-serve | [bond.md](https://agentlist.wiki/agents/bond.md) |
| Catch | Catch | Waitlist | US$99/mo; 10-day trial | Provider cloud | None described | [catch.md](https://agentlist.wiki/agents/catch.md) |
| Claude Code | Anthropic | Available | Subscription or API usage | Terminal, IDE, desktop, and cloud sessions | unknown | [claude-code.md](https://agentlist.wiki/agents/claude-code.md) |
| Claude Duet | Anthropic | Waitlist · some regions, paid plan | Claude Pro $20/mo | Provider cloud | Self-serve | [claude-duet.md](https://agentlist.wiki/agents/claude-duet.md) |
| Codex | OpenAI | Available · some regions, paid plan (limited) | From $20/mo plan; API option | Local or managed cloud | unknown | [codex.md](https://agentlist.wiki/agents/codex.md) |
| Cue | Manus AI | Access unverified (limited) | unknown | Unverified (limited) | unknown | [cue.md](https://agentlist.wiki/agents/cue.md) |
| Eney | MacPaw | Available | Trial, then $14.99/mo | Local + cloud models | Self-serve | [eney.md](https://agentlist.wiki/agents/eney.md) |
| Gemini Tasks | Google | Available · paid plan, age limits | AI Pro $19.99/mo | Google Cloud VMs | Self-serve | [gemini-spark.md](https://agentlist.wiki/agents/gemini-spark.md) |
| Grok Bot | Grok / Cursor | Available | From US$30/mo | Provider cloud | None described | [grok-bot.md](https://agentlist.wiki/agents/grok-bot.md) |
| Hark Pro | Hark Labs | Available | Free; $20–$100/mo | Provider cloud | On request | [hark.md](https://agentlist.wiki/agents/hark.md) |
| Hermes Agent | Nous Research | Available | Free; you pay models | Self-hosted | Partial | [hermes.md](https://agentlist.wiki/agents/hermes.md) |
| Instinct | Spear Street Technology, Inc. | Access unverified (limited) | unknown | Unverified (limited) | None described | [instinct.md](https://agentlist.wiki/agents/instinct.md) |
| IronClaw | NEAR AI | Available | Free; you pay models | Self-hosted | Partial | [ironclaw.md](https://agentlist.wiki/agents/ironclaw.md) |
| Khoj | Khoj AI | Available | Free; no hosted tier | Self-hosted | Partial (limited) | [khoj.md](https://agentlist.wiki/agents/khoj.md) |
| Kimi Claw | Kimi | Available · paid plan | CNY199/mo (China) | Kimi cloud or self | On request | [kimi-claw.md](https://agentlist.wiki/agents/kimi-claw.md) |
| Letta | Letta | Available | Free; Cloud from US$20/mo | Self-hosted or Letta Cloud | Self-serve | [letta.md](https://agentlist.wiki/agents/letta.md) |
| Lindy | Lindy | Available | $49.99/user/mo | Provider cloud | On request | [lindy.md](https://agentlist.wiki/agents/lindy.md) |
| Lucas | Lucas AI | Available | Free; $20 or $100/mo | Provider cloud | Self-serve (limited) | [lucas.md](https://agentlist.wiki/agents/lucas.md) |
| Manus | Manus AI | Available | Free; from US$20/mo | Provider cloud + your computer | Self-serve | [manus.md](https://agentlist.wiki/agents/manus.md) |
| Martin | DCWU, Inc. | Available | From US$21/mo billed yearly | Provider cloud | unknown | [martin.md](https://agentlist.wiki/agents/martin.md) |
| MaxClaw | MiniMax | Available | $19/mo | Provider cloud | On request | [maxclaw.md](https://agentlist.wiki/agents/maxclaw.md) |
| Moltis | Fabien Penso | Available | Free; you pay models | Self-hosted | Self-serve | [moltis.md](https://agentlist.wiki/agents/moltis.md) |
| Muse | Meta | Available · some regions (limited) | Free; paid upgrades (limited) | Provider cloud | Self-serve | [muse.md](https://agentlist.wiki/agents/muse.md) |
| nanobot | HKUDS | Available | Free; you pay models | Self-hosted | Partial | [nanobot.md](https://agentlist.wiki/agents/nanobot.md) |
| NanoClaw | NanoCo | Available | Free; you pay models | Self-hosted | Partial | [nanoclaw.md](https://agentlist.wiki/agents/nanoclaw.md) |
| Noah | Hey Noah AI Inc. | Access unverified · age limits | Not published (limited) | Provider cloud | On request | [heynoah.md](https://agentlist.wiki/agents/heynoah.md) |
| Omi | Based Hardware Inc. | Available | Free; Pro US$269/yr | Hosted, or self-host the backend | On request | [omi.md](https://agentlist.wiki/agents/omi.md) |
| OpenAI Dots | OpenAI | Available · rolling out, paid plan, some regions | ChatGPT Pro $100/mo | Cloud + your PC | Self-serve | [dots.md](https://agentlist.wiki/agents/dots.md) |
| OpenClaw | OpenClaw Foundation | Available | Free; you pay models | Self-hosted | Partial | [openclaw.md](https://agentlist.wiki/agents/openclaw.md) |
| OpenFang | RightNow AI | Available | Free; you pay models | Self-hosted | Partial | [openfang.md](https://agentlist.wiki/agents/openfang.md) |
| OpenHuman | Tiny Human Intelligence Inc | Available | Free app; optional inference tiers | Your desktop or your server | Partial (limited) | [openhuman.md](https://agentlist.wiki/agents/openhuman.md) |
| Pally | Pally Technologies | Available | Free; paid Cloud Mac | unknown | Self-serve | [pally.md](https://agentlist.wiki/agents/pally.md) |
| Perplexity Computer | Perplexity AI, Inc. | Available · paid plan | Pro US$20/mo; Max US$200/mo | Provider cloud, or your hardware | Self-serve | [perplexity-computer.md](https://agentlist.wiki/agents/perplexity-computer.md) |
| Persona | Iris Assistant, Inc. | Available | Free; $179 Band | Provider cloud | On request | [persona.md](https://agentlist.wiki/agents/persona.md) |
| Pickle | Pickle, Inc. | Available · age limits | From $20/mo | Provider cloud | On request | [pickle.md](https://agentlist.wiki/agents/pickle.md) |
| PicoClaw | Sipeed | Available | Free; you pay models | Self-hosted | Partial | [picoclaw.md](https://agentlist.wiki/agents/picoclaw.md) |
| Poke | The Interaction Company | Available | Free; $19 or $199/mo | Provider cloud | On request | [poke.md](https://agentlist.wiki/agents/poke.md) |
| QwenPaw | AgentScope | Available | Free; free hosted option | Self-hosted or free cloud | Self-serve | [qwenpaw.md](https://agentlist.wiki/agents/qwenpaw.md) |
| Rene | Marumaru / Second Enlightenment | Available | Free to start | Provider cloud | On request | [rene.md](https://agentlist.wiki/agents/rene.md) |
| Simular | Simular Inc. | Waitlist | US$50–500/mo | Your machine or its cloud | On request | [simular.md](https://agentlist.wiki/agents/simular.md) |
| Skye | Signull Labs | Available · some regions | Free | Provider cloud | On request | [skye.md](https://agentlist.wiki/agents/skye.md) |
| Thine | Foyer Tech | Available | Free; $14.99/mo | Provider cloud | None described | [thine.md](https://agentlist.wiki/agents/thine.md) |
| Tomo | Mapo Labs, Inc. | Available | From $19.99/mo | Provider cloud | None described | [tomo.md](https://agentlist.wiki/agents/tomo.md) |
| Town | Town.com, Inc. | Available | Free; $15–$49/mo | Provider cloud | On request | [town.md](https://agentlist.wiki/agents/town.md) |
| Underdog | Conway Research | Available · invite | Free | On your device | Partial (limited) | [underdog.md](https://agentlist.wiki/agents/underdog.md) |
| Vora Aegis | Vora | Reservation · some regions | US$2,499 device | Your own appliance | Self-serve | [vora.md](https://agentlist.wiki/agents/vora.md) |
| Wabi | Wabi, Inc. | Available | Free download (limited) | Cloud + Mac context | On request | [wabi.md](https://agentlist.wiki/agents/wabi.md) |
| ZeroClaw | ZeroClaw Labs | Available | Free; you pay models | Self-hosted | Partial | [zeroclaw.md](https://agentlist.wiki/agents/zeroclaw.md) |
| Zo Computer | Zo Computer Company | Available | US$18–200/mo after trial | Zo cloud, plus your machine | On request | [zo.md](https://agentlist.wiki/agents/zo.md) |

Fields, states and evidence rules: https://agentlist.wiki/method.md. Structured dataset: https://agentlist.wiki/data.json.

---

# Agent Zero

> An agent framework that gives the agent its own Dockerised Linux computer with a terminal, desktop and browser, and can also bridge to the machine you run it on.

- **Maker:** Agent Zero, s.r.o.
- **Catalogue type:** Self-hosted runtime
- **Website:** https://agent-zero.ai
- **Source code:** https://github.com/agent0ai/agent-zero
- **Also known as:** AgentZero, A0
- **Evidence:** 12 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/agent-zero/
- **This record:** https://agentlist.wiki/agents/agent-zero.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 4 |
| Price & usage costs | Free; you pay models | documented | 1, 2 |
| Included in another plan | Open-source software with no hosted tier | documented | 1 |
| Where it runs | Self-hosted | documented | 4 |
| Open source & license | MIT | documented | 3 |
| Controls your computer | Yes · Mac, Windows, Linux | documented | 9 |
| Its own computer | Own hardware · Watch and take over | documented | 7, 8 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 3, 4 |
| Remote shut-off | Not possible | documented | 4 |
| What you keep | Partial | limited | 2, 5 |
| Company type | Independent project | documented | 1, 3 |
| Launch date | Jun 26, 2024 | documented | 12 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Independent project [1, 3]
- **Company stage** `documented` — Agent Zero, s.r.o., Brno, Czech Republic [1, 3]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Jun 26, 2024 · Open-source release · First public release [12]

**Launch history**

- 2024-06-26 · Open-source release · First public release [12]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — The agent gets its own Linux computer inside Docker, with a terminal, an XFCE desktop and a browser it can watch and drive, plus Time Travel snapshots you can revert [2, 8]
- **Computer use** `documented` — Two separate scopes: an XFCE desktop inside its own Docker container, and the A0 CLI Connector bridging to the host's files, shell and browser [8, 9]
- **Controls your computer** `documented` — The A0 CLI Connector bridges the agent to the host's files, shell and browser. [9]
- **Its own computer** `documented` — An XFCE Linux desktop inside its Docker container, which the user can watch and drive in a canvas panel. [7, 8]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — A built-in browser in its container, observable in a canvas panel; the CLI connector can also reach the host browser [7, 9]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Tasks run scheduled, planned or ad hoc, scoped to a project, with a notification system [6]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — The A0 Launcher installs on macOS, Linux and Windows, on x86 and ARM; the agent interface itself is web [4]
- **iPhone & Android apps** `documented` — No mobile app [2]
- **Web & messaging** `documented` — Web interface, the A0 Launcher desktop app and a command-line interface [2, 4]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — HTTP API with a key from settings, covering messaging, logs, chat reset and files, plus MCP and A2A [11]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Model presets for main, utility and embedding roles with your own provider keys; an OpenAI Codex plan connects over OAuth [10]
- **Bring your own model** `documented` — Bring your own provider keys; local models are documented for tool use [10]
- **Included in another plan** `documented` — Open-source software with no hosted tier [1]
- **Price & usage costs** `documented` — Free software; no hosted service or paid tier is published [1, 2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — MIT-licensed software you run yourself in Docker or from a desktop installer, so it keeps running if the company stops. [3, 4]
- **Remote shut-off** `documented` — Runs on your own hardware with no activation or licence check described. [4]
- **What you keep** `limited` (partial or conditional only) — State lives in a Docker volume you control and memory entries can be copied individually, but no single export command is described. [2, 5]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT. GitHub reports no assertion because contact lines precede the licence body, but the text is verbatim MIT [3]
- **Where it runs** `documented` — Docker on your own machine, an installer for macOS, Linux and Windows, or a VPS you control [4]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Time Travel keeps snapshot history with diffs so a run can be reverted [2]
- **What it remembers** `documented` — Facts, solutions, preferences and imported knowledge persist, with a dashboard to search, read, copy, edit and delete individual entries [5]
- **Edit & export your data** `limited` (partial or conditional only) — Memory entries can be copied individually from the dashboard; all state lives in a Docker volume you control, though no single export command is described [2, 5]
- **Privacy & isolation** `documented` — Tool execution is contained in its own Docker container, separate from the host unless the CLI connector is used [4, 9]
- **Availability & access** `documented` — Open source; install by Docker, a desktop launcher or a script [4]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | US$0 | Free software | Full features when self-hosted; Your own provider keys | You provide hosting and model costs | 1, 10 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | The agent interface runs in a browser against your own instance. | 4 |
| macOS | Available | A0 Launcher desktop installer for macOS, x86 and ARM. | 4 |
| Windows | Available | A0 Launcher desktop installer for Windows. | 4 |
| Linux | Available | A0 Launcher desktop installer for Linux. | 4 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-13 · Release · [Agent Zero v2.14](https://github.com/agent0ai/agent-zero/releases/tag/v2.14) · Agent Zero — Startup and chat loading are significantly faster, with embedding warm-up overlapping chat restoration for roughly 27% lower median time to first token, alongside upgraded model-routing dependencies.

## Sources

1. [Agent Zero official website](https://agent-zero.ai) — checked 2026-10-22
2. [Agent Zero repository](https://github.com/agent0ai/agent-zero) — checked 2026-10-22
3. [Agent Zero licence](https://github.com/agent0ai/agent-zero/blob/main/LICENSE) — checked 2026-10-22
4. [Agent Zero installation](https://github.com/agent0ai/agent-zero/blob/main/docs/setup/installation.md) — checked 2026-10-22
5. [Agent Zero memory dashboard](https://github.com/agent0ai/agent-zero/blob/main/docs/guides/memory.md) — checked 2026-10-22
6. [Agent Zero tasks and scheduling](https://github.com/agent0ai/agent-zero/blob/main/docs/guides/usage.md) — checked 2026-10-22
7. [Agent Zero browser](https://github.com/agent0ai/agent-zero/blob/main/docs/guides/browser.md) — checked 2026-10-22
8. [Agent Zero Linux desktop](https://github.com/agent0ai/agent-zero/blob/main/docs/guides/desktop.md) — checked 2026-10-22
9. [Agent Zero CLI connector](https://github.com/agent0ai/agent-zero/blob/main/docs/guides/a0-cli-connector.md) — checked 2026-10-22
10. [Agent Zero model presets](https://github.com/agent0ai/agent-zero/blob/main/docs/guides/model-presets.md) — checked 2026-10-22
11. [Agent Zero connectivity and API](https://github.com/agent0ai/agent-zero/blob/main/docs/developer/connectivity.md) — checked 2026-10-22
12. [Agent Zero first release](https://github.com/agent0ai/agent-zero/releases/tag/v0.2) — checked 2026-10-22

---

# Alexa+

> Amazon's generative Alexa, with remembered personal details, routines and agentic web tasks across Echo devices, the Alexa app and the web.

- **Maker:** Amazon
- **Catalogue type:** Hosted agent
- **Website:** https://www.amazon.com/alexaplus
- **Source code:** unknown — no public repository recorded
- **Also known as:** Alexa Plus, Amazon Alexa
- **Evidence:** 6 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/alexa-plus/
- **This record:** https://agentlist.wiki/agents/alexa-plus.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · some regions | documented | 3 |
| Price & usage costs | Free with Prime; US$19.99/mo | documented | 2, 3 |
| Included in another plan | Included with Prime | documented | 3 |
| Where it runs | Provider cloud | documented | 1, 3 |
| Open source & license | unknown | unknown | — |
| Controls your computer | No | documented | 2 |
| Its own computer | Cloud · no remote view | documented | 3 |
| Works while your computer is off | Keeps working | limited | 3 |
| If the vendor stops | Stops | limited | 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 6 |
| Company type | Public company | documented | 1 |
| Launch date | Feb 26, 2025 | documented | 2 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Public company [1]
- **Company stage** `documented` — Public company · NASDAQ: AMZN [1]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 26, 2025 · Announcement · Announced at Amazon's Devices and Services event [2]

**Launch history**

- 2025-02-26 · Announcement · Announced at Amazon's Devices and Services event [2]
- 2026 · Public launch · General availability in the US and Canada, free for Prime [3]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Reaches Amazon's own services, including Ring footage and AI video search, Prime Video, Amazon Music and Photos, alongside shopping and smart home [3, 4]
- **Computer use** `limited` (partial or conditional only) — Navigates the internet in a self-directed way to complete tasks behind the scenes; no control of the user's own device is described [2]
- **Controls your computer** `documented` — Nothing runs on the user's computer; Alexa+ acts from Amazon's side and on Echo hardware. [2]
- **Its own computer** `documented` — Runs in Amazon's cloud. No remote view of a machine of its own is offered. [3]
- **Works while your computer is off** `limited` (partial or conditional only) — Runs on Amazon's cloud and Echo hardware rather than the user's computer, so tasks do not depend on a personal device being on. [3]
- **Browser control** `documented` — Navigates the web on the user's behalf from Amazon's side [2]
- **Voice & ambient listening** `documented` — Voice is the primary interface, across Echo devices, Echo Show, Fire TV and the Alexa app [1]
- **Background work & triggers** `limited` (partial or conditional only) — Routines, reminders and calendar management, with agentic tasks completed behind the scenes; a user-facing scheduler for autonomous runs is not described [1, 2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — No native desktop app; the web experience is at Alexa.com [1]
- **iPhone & Android apps** `documented` — Alexa apps for iPhone and Android [1]
- **Web & messaging** `documented` — Echo devices, Echo Show, Echo Hub, Fire TV, the Alexa app and Alexa.com [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `documented` — Runs on Echo speakers, Echo Show, Echo Hub and Fire TV, and is also reachable from the Alexa app and Alexa.com without buying hardware. [1, 3]
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Amazon Nova and Anthropic models run in Amazon's cloud [3]
- **Bring your own model** `limited` (partial or conditional only) — Fully managed; personal model keys are not described [3]
- **Included in another plan** `documented` — Included with Prime at no extra cost; otherwise a standalone subscription [3]
- **Price & usage costs** `documented` — US$19.99 a month, free for Prime members across the whole household; non-Prime customers can also use a limited free text chat on Alexa.com and in the app [2, 3]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted and tied to Amazon's services and devices. The company has not published what happens if it is discontinued. [3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Amazon's general request-your-personal-information flow returns account data by download link; no self-serve export of remembered items is described. [6]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Amazon's cloud, reached through Echo devices, the app and the web [1, 3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Remember This stores details on request; remembered items are reviewable under Alexa Privacy, and Amazon notes they may be accessible to anyone using the household's devices [5]
- **Edit & export your data** `limited` (partial or conditional only) — Amazon's general personal-information request covers account data; a dedicated export of remembered items is not described [6]
- **Privacy & isolation** `documented` — Voice recordings, transcripts and typed requests can be reviewed and deleted in Alexa Privacy settings [5]
- **Availability & access** `documented` — Generally available in the United States and Canada, with early access in Brazil, the United Kingdom, Mexico, Italy, Spain, Germany, Austria and France [3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Included with Prime | US$0 | Included in a Prime membership | Full Alexa+ across every Alexa-enabled device, Alexa.com and the app; Covers the whole household | Requires a Prime membership | 3 |
| Alexa+ | US$19.99/month | Monthly, for customers without Prime | Full Alexa+ experience across devices, web and the app | A limited free text chat is available without paying | 2, 3 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Alexa.com, plus a limited free text chat for customers without Prime. | 3 |
| iOS | Available | Alexa app for iPhone. | 1 |
| Android | Available | Alexa app for Android. | 1 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`optional` Optional hardware — **Echo devices**: Runs on Echo speakers, Echo Show, Echo Hub and Fire TV, and is also reachable from the Alexa app and Alexa.com without buying hardware. [1, 3]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.aboutamazon.com — No dedicated changelog; updates appear as dated newsroom posts (checked 2026-10-22)

- 2026-10-08 · Release · [Amazon Alexa Tablets with Alexa+ built in](https://www.aboutamazon.com/news/devices/new-amazon-alexa-tablets-alexa-plus) · Amazon — Three tablets at 8, 11 and 12 inches ship with Alexa+ and on-screen intelligence built in, from US$229.99.
- 2026-09-30 · Release · [Alexa+ comes to Fire TV with conversational search and a hub](https://www.aboutamazon.com/news/devices/alexa-plus-fire-tv-hub-mobile-app-updates) · Amazon — Adds complex conversational search on Fire TV and a dedicated Alexa+ hub carrying history across devices.
- 2026-08-19 · Availability change · [Alexa+ free on Fire TV for all US customers](https://www.aboutamazon.com/news/devices/alexa-plus-fire-tv-free-ai) · Amazon — Eligible Fire TV devices in the United States are upgraded to Alexa+ with no subscription and no Prime requirement.

## Sources

1. [Alexa+ overview](https://www.aboutamazon.com/what-we-do/devices-services/alexa-plus) — checked 2026-10-22
2. [Amazon introduces the next generation of Alexa](https://www.aboutamazon.com/news/devices/new-alexa-generative-artificial-intelligence) — checked 2026-10-22
3. [Alexa+ is available free for Prime members](https://www.aboutamazon.com/news/devices/alexa-plus-available-free-prime-members-us) — checked 2026-10-22
4. [Alexa+ and Ring doorbell greetings](https://www.aboutamazon.com/news/devices/alexa-ring-doorbell-ai-greetings) — checked 2026-10-22
5. [Alexa privacy settings](https://www.amazon.com/alexaprivacysettings) — checked 2026-10-22
6. [Request your personal information from Amazon](https://www.amazon.com/gp/help/customer/display.html?nodeId=TP1zlemejtTn6pwYKS) — checked 2026-10-22

---

# Ambi

> Ambient intelligence carrying context, memory, tasks, and routines across devices and connected tools.

- **Maker:** Ambi
- **Catalogue type:** Hosted agent
- **Website:** https://www.ambi.ai/en
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 6 sources, checked 2026-10-21 to 2026-10-26
- **Profile:** https://agentlist.wiki/agent/ambi/
- **This record:** https://agentlist.wiki/agents/ambi.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1, 3 |
| Price & usage costs | Trial, then paid | documented | 2 |
| Included in another plan | Standalone Ambi Plus, Pro or Max subscription | documented | 2 |
| Where it runs | Unverified | limited | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | Limited | limited | 1 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 5 |
| Company type | Startup | documented | 1 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [1]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Turns live conversations into Action Spotter follow-ups, Tasks and ongoing Routines across devices [2]
- **Computer use** `limited` (partial or conditional only) — Reads permitted conversations, screens and connected tools; full desktop control is not specified [1]
- **Controls your computer** `limited` (partial or conditional only) — Reads permitted conversations, screens and connected tools; full desktop control is not specified. [1]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Real-time ambient Listening, with visible on/off control; transcription, recaps and follow-up [2]
- **Background work & triggers** `documented` — Routines follow a standing brief and keep working as new context arrives [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — macOS on Apple silicon; Windows 11 x64 [3]
- **iPhone & Android apps** `documented` — iPhone and Apple Watch; Android is coming soon [3, 6]
- **Web & messaging** `documented` — macOS, Windows, iOS and watch context; connected messaging and work tools [1, 2, 3]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `limited` (partial or conditional only) — Dedicated hardware is announced for early 2027; the public site is accepting waitlist registrations. [1]
- **Developer API** `limited` (partial or conditional only) — MCP connections on Pro and Max; public developer API not specified [2]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `documented` — Standalone Ambi Plus, Pro or Max subscription [2]
- **Price & usage costs** `documented` — Seven-day trial: 4,000 credits and 3 full Insights; paid plans use shared weekly credits [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A right to receive a copy of held personal information in a portable format on request; no self-serve export is described. Confirmed deletion removes account information and stored content immediately. [5]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `limited` (partial or conditional only) — Connected software and hardware; execution architecture unverified [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Actions that cannot be undone wait for your approval [1, 2]
- **What it remembers** `documented` — Personal memory built across conversations and devices, under user control [2]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — Permission-based context access; decisions remain with user [1]
- **Availability & access** `documented` — Software downloads listed for macOS, Windows and iOS; hardware expected early next year [1, 3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free trial | US$0 | One seven-day trial | 4,000 credits, 3 active Routines and 3 full Insights; Listening, transcription, Recaps and connectors | No card required; no automatic charge; Trial credits expire after seven days | 4 |
| Plus | Not published | Monthly or yearly; yearly saves 15% | Everyday use with shared weekly credits; User-started recording and transcription do not consume credits | Weekly credits do not roll over; Paid top-ups last twelve months | 4 |
| Pro | Not published | Monthly or yearly; yearly saves 15% | More active Routines than Plus; External MCP connections | Shared credits reset weekly; Top-ups do not increase Routine or Insight limits | 4 |
| Max | Not published | Monthly or yearly; yearly saves 15% | Largest weekly credit allowance; Unlimited Routines and nightly Insights | Credit-using features pause when the balance runs out; Recording and transcription remain available | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | macOS on Apple silicon; Windows 11 x64 | 3 |
| Windows | Available | macOS on Apple silicon; Windows 11 x64 | 3 |
| iOS | Available | iOS; Android is coming soon | 3 |
| Android | Coming soon | iOS; Android is coming soon | 3 |
| Apple Watch | Available | The App Store listing supports Apple Watch alongside iPhone. | 6 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`announced` Announced, not shipping — **Ambi hardware**: Dedicated hardware is announced for early 2027; the public site is accepting waitlist registrations. [1]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Ambi early access and hardware announcement](https://www.ambi.ai/en) — checked 2026-10-21
2. [Ambi features: Listening, Memory, Tasks and Routines](https://www.ambi.ai/en/features) — checked 2026-10-21
3. [Download Ambi for desktop and mobile](https://www.ambi.ai/download/) — checked 2026-10-21
4. [Ambi plans, trial and credit rules](https://www.ambi.ai/pricing/) — checked 2026-10-21
5. [Ambi privacy policy](https://ambi.ai/en/privacy) — checked 2026-10-22
6. [Ambi: Ambient intelligence on the App Store](https://apps.apple.com/us/app/ambi-ambient-intelligence/id6787026123) — checked 2026-10-26

---

# Bond

> A work assistant that reads connected tools and keeps one prioritized to-do list of commitments, drafting follow-ups and briefs for the person to send.

- **Maker:** BOND App Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://www.bondapp.io
- **Source code:** unknown — no public repository recorded
- **Also known as:** BOND, Bond App
- **Evidence:** 6 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/bond/
- **This record:** https://agentlist.wiki/agents/bond.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2, 4 |
| Price & usage costs | Free; US$49/mo | documented | 2 |
| Included in another plan | Standalone Bond individual or Enterprise subscription | documented | 2 |
| Where it runs | Provider cloud | documented | 1, 3 |
| Open source & license | unknown | unknown | — |
| Controls your computer | No | limited | 1, 2 |
| Its own computer | None · no remote view | limited | 1, 2 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1, 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 2 |
| Company type | Startup | documented | 5 |
| Launch date | May 28, 2025 | documented | 6 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [5]
- **Company stage** `documented` — Y Combinator Spring 2025 [5]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — May 28, 2025 · Public launch · Launch YC [6]

**Launch history**

- 2025-05-28 · Public launch · Launch YC [6]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Pulls commitments out of connected work tools into one self-updating to-do list ranked P0–P3, with a morning brief and per-meeting context briefs [1, 2]
- **Computer use** `limited` (partial or conditional only) — Reads and writes through connected app integrations; no control of your computer or of a cloud desktop is described [1, 2]
- **Controls your computer** `limited` (partial or conditional only) — Nothing runs on your machine; it works only through connected app integrations. [1, 2]
- **Its own computer** `limited` (partial or conditional only) — No machine of its own. It reads and writes through connected accounts rather than operating a computer. [1, 2]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Researches meeting context from connected tools; interactive browser control is not specified [1, 2]
- **Voice & ambient listening** `limited` (partial or conditional only) — Ingests notes from connected meeting recorders such as Granola, Fireflies, Fathom, Circleback and Read AI; its own voice interface or ambient listening is not specified [2]
- **Background work & triggers** `documented` — Runs a daily morning brief, prepares briefs ahead of meetings, tracks overdue commitments owed by others, and runs custom routines on a daily, weekly or monthly cycle [1, 2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — No native desktop app; the product states there is no new app to learn and runs in Slack and the web application [1, 4]
- **iPhone & Android apps** `documented` — No native iPhone or Android app; mobile access is through iMessage and Slack [1, 2]
- **Web & messaging** `documented` — Web application, Slack and iMessage [1, 2, 4]
- **Messaging services** `documented` — iMessage, Slack [1, 2, 4]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — No public API is specified; Enterprise lists custom integrations and deployment options arranged with the team [2]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — Third-party AI model providers under zero-data-retention agreements; the specific models and providers are not named [2, 3]
- **Bring your own model** `limited` (partial or conditional only) — Fully managed service; personal model keys or model selection are not specified [2]
- **Included in another plan** `documented` — Standalone Bond individual or Enterprise subscription [2]
- **Price & usage costs** `documented` — Free tier with 10,000 credits; individual plan US$49/month; Enterprise priced on request [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1, 3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Everything can be exported, and the company states it deletes customer data within 30 days of cancellation. [2]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Provider-hosted service that connects to the user's third-party work accounts [1, 3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Drafts follow-ups, replies and briefs and leaves sending to the person; the product describes the user staying in control of what gets sent [1]
- **What it remembers** `documented` — Builds lasting company context from documents, messages and decisions in connected tools; each person keeps their own context and to-do list [1, 2]
- **Edit & export your data** `documented` — Everything can be exported, and data is deleted within 30 days of cancellation [2]
- **Privacy & isolation** `documented` — Zero-data-retention agreements with model providers and no training on customer data; encrypted in transit and at rest; continuously monitored against ISO 27001 controls through Aikido Security. SOC 2 Type II, SSO, SCIM and audit logs are listed under Enterprise [1, 2]
- **Availability & access** `documented` — Public sign-up with a free credit allowance; Enterprise access is arranged with the team [2, 4]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free tier | 10,000 credits; Self-managing to-do list, follow-up and commitment tracking; Slack and iMessage access; all Bond integrations | Credit allowance is one-off; renewal terms are not published | 2 |
| Bond | US$49/month | Stated as a monthly price; annual billing terms are not published | Self-managing to-do list that updates while you work; Automatic follow-up and commitment tracking; Available in Slack and iMessage; Automated routines for repetitive work; All Bond integrations included | Included credit allowance is not published | 2 |
| Enterprise | Custom | Priced on request | Shared company brain across the organization; SOC 2 Type II, SSO, SCIM and audit logs; Dedicated onboarding and a forward-deployed AE; Custom integrations and deployment options; SLA and priority support | Requires contacting the team; no published price | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web application at app.bondapp.io, with Google, Microsoft or work-email sign-in | 4 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Web application, Slack and iMessage | 1, 2, 4 |
| Slack | Available | Web application, Slack and iMessage | 1, 2, 4 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Bond official website](https://www.bondapp.io/) — checked 2026-10-21
2. [Bond pricing, plans and frequently asked questions](https://www.bondapp.io/pricing) — checked 2026-10-21
3. [Bond privacy policy](https://www.bondapp.io/privacy-policy) — checked 2026-10-21
4. [Bond web application sign-in](https://app.bondapp.io/login) — checked 2026-10-21
5. [Bond company profile, Y Combinator](https://www.ycombinator.com/companies/bond) — checked 2026-10-21
6. [Launch YC: BOND, AI Chief of Staff for CEOs and busy execs](https://www.ycombinator.com/launches/Ncl-bond-ai-chief-of-staff-for-ceos-and-busy-execs) — checked 2026-10-21

---

# Catch

> An AI executive assistant for scheduling, inbox triage, travel and outbound calls, reached through ordinary chat and email rather than an app. Closed beta.

- **Maker:** Catch
- **Catalogue type:** Hosted agent
- **Website:** https://www.catchagent.ai
- **Source code:** unknown — no public repository recorded
- **Also known as:** Catch Agent
- **Evidence:** 3 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/catch/
- **This record:** https://agentlist.wiki/agents/catch.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Waitlist | documented | 1 |
| Price & usage costs | US$99/mo; 10-day trial | documented | 1 |
| Included in another plan | Standalone Catch subscription | documented | 1 |
| Where it runs | Provider cloud | documented | 3 |
| Open source & license | unknown | unknown | — |
| Controls your computer | No | limited | 1 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | None described | documented | 3 |
| Company type | Startup | documented | 2 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [2]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — No app to manage: reached through Slack, Messages, email or a phone call, and a scheduling thread can be handed over simply by copying in an address, free and without an account [1]
- **Computer use** `limited` (partial or conditional only) — Acts through integrations and outbound phone calls from its own cloud; no control of a browser or of the user's machine is described [1]
- **Controls your computer** `limited` (partial or conditional only) — Nothing runs on the user's machine; Catch acts through integrations and phone calls from its own cloud. [1]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Browser control is not described; form filling and booking are presented as agent actions awaiting approval [1]
- **Voice & ambient listening** `documented` — Makes outbound phone calls, and is reachable by call [1]
- **Background work & triggers** `documented` — Described as proactive by default, with a morning brief, daily sync, smart reminders, proactive updates, follow-ups and calendar protection [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — No desktop app; the product states there is no app to manage [1]
- **iPhone & Android apps** `documented` — No mobile app; the product states there is no app to manage [1]
- **Web & messaging** `documented` — Slack, Messages, email and phone calls, with integrations across more than seventy tools [1]
- **Messaging services** `documented` — iMessage, SMS, WhatsApp, Slack, Microsoft Teams, Email [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — Described as built on the company's own AI, while the security pages disclose unnamed third-party vendors contractually barred from retaining or training on customer data [1, 3]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `documented` — Standalone Catch subscription [1]
- **Price & usage costs** `documented` — US$99 a month, billed monthly and cancellable at any time, with a ten-day free trial. One scheduling feature is free and needs no account [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted with no self-hosting option. The company has not published what happens if it shuts down. [3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Access can be revoked and data deleted on request, but no export or portability right is described. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Provider cloud, with AWS Secrets Manager and key management named; no self-hosting [3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `limited` (partial or conditional only) — Described as adapting to your rhythm, priorities and preferences, with OAuth-scoped access to email and calendar; no memory viewer or editor is described [1, 3]
- **Edit & export your data** `limited` (partial or conditional only) — No export is described; the security pages offer revocation of access and deletion on request [3]
- **Privacy & isolation** `documented` — SOC 2 Type II and CASA Tier-2 are listed, and AI vendors are contractually barred from retaining or training on customer data [3]
- **Availability & access** `documented` — Closed beta with limited spots, entered by answering a short eligibility questionnaire or joining a waiting list [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Catch | US$99/month | Billed monthly, cancel at any time; ten-day free trial | Works across all your calendars; Schedules meetings and drafts email in your voice; Outgoing phone calls; Unlimited tasks, reminders and integrations | Closed beta: a spot must be available | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

unknown — no native app or web client is recorded.

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Reachable through Messages; iMessage appears in the integration list. | 1 |
| SMS | Available | Listed among Catch's integrations. | 1 |
| WhatsApp | Available | Listed among Catch's integrations. | 1 |
| Slack | Available | One of the four ways to chat with Catch. | 1 |
| Microsoft Teams | Available | Listed among Catch's integrations. | 1 |
| Email | Available | Reachable by email, including copying it on a scheduling thread. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Catch official website](https://www.catchagent.ai) — checked 2026-10-22
2. [About Catch](https://www.catchagent.ai/about) — checked 2026-10-22
3. [Catch security](https://www.catchagent.ai/security) — checked 2026-10-22

---

# Claude Code

> A coding harness with memory, desktop scheduling, and cloud sessions. Tracked as adjacent because its primary scope is software work.

- **Maker:** Anthropic
- **Catalogue type:** Coding tool (adjacent)
- **Website:** https://code.claude.com/docs/en/overview
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 11 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/claude-code/
- **This record:** https://agentlist.wiki/agents/claude-code.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Subscription or API usage | documented | 8 |
| Included in another plan | Claude Pro / Max; API option | documented | 8 |
| Where it runs | Terminal, IDE, desktop, and cloud sessions | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | unknown | unknown | — |
| Company type | Major AI lab | documented | 10 |
| Launch date | Feb 24, 2025 | documented | 11 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Major AI lab [10]
- **Company stage** `documented` — Series H [10]
- **Disclosed funding** `documented` — US$65B · Series H · May 28, 2026 [10]
- **Launch date** `documented` — Feb 24, 2025 · Research preview [11]

**Launch history**

- 2025-02-24 · Research preview [11]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Works directly in a codebase; can be scripted in CI or controlled from a phone. [5, 6]
- **Computer use** `limited` (partial or conditional only) — Local desktop control: macOS and Windows through Desktop, macOS through CLI; CLI requires Pro/Max and interactive use. [2]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Chrome extension for browser automation. [1]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `limited` (partial or conditional only) — Desktop recurring tasks; configuration dependent [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native Claude desktop app for macOS and Windows; Linux desktop is in beta. [3, 4]
- **iPhone & Android apps** `documented` — Claude apps for iOS and Android can start and steer cloud or remotely connected sessions; code runs elsewhere. [5]
- **Web & messaging** `documented` — Terminal, IDE, desktop, browser, mobile [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Agent SDK for Python and TypeScript, plus non-interactive CLI execution. [6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Claude models via subscription, API, or supported cloud providers. [7]
- **Bring your own model** `documented` — Claude subscription or own Anthropic API credentials; supported cloud providers include Amazon Bedrock, Google Cloud, and Microsoft Foundry. [1, 7, 8]
- **Included in another plan** `documented` — Included with eligible Claude subscriptions, including Pro and Max; shared plan limits or separate API billing. [8]
- **Price & usage costs** `documented` — Eligible Claude subscription usage or pay-as-you-go API billing. [8]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Terminal, IDE, desktop, and cloud sessions [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Configurable permission rules; manual approvals or automatic action review. [9]
- **What it remembers** `documented` — Project instructions and cross-session auto memory [1]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `documented` — Multiple surfaces with account requirements [1]

## Plans

unknown — no published plan or price list is recorded. See the Price & usage costs claim above.

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Terminal, IDE, desktop, browser, mobile | 1 |
| macOS | Available | Native Claude desktop app for macOS and Windows; Linux desktop is in beta. | 3, 4 |
| Windows | Available | Native Claude desktop app for macOS and Windows; Linux desktop is in beta. | 3, 4 |
| Linux | Beta | Native Claude desktop app for macOS and Windows; Linux desktop is in beta. | 3, 4 |
| iOS | Available | Claude apps for iOS and Android can start and steer cloud or remotely connected sessions; code runs elsewhere. | 5 |
| Android | Available | Claude apps for iOS and Android can start and steer cloud or remotely connected sessions; code runs elsewhere. | 5 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Claude Code overview](https://code.claude.com/docs/en/overview) — checked 2026-10-21
2. [Let Claude use your computer](https://code.claude.com/docs/en/computer-use) — checked 2026-10-21
3. [Claude Code desktop application](https://code.claude.com/docs/en/desktop) — checked 2026-10-21
4. [Claude Desktop on Linux (beta)](https://code.claude.com/docs/en/desktop-linux) — checked 2026-10-21
5. [Claude Code on mobile](https://code.claude.com/docs/en/mobile) — checked 2026-10-21
6. [Run Claude Code programmatically](https://code.claude.com/docs/en/headless) — checked 2026-10-21
7. [Claude Code deployment options](https://code.claude.com/docs/en/third-party-integrations) — checked 2026-10-21
8. [Use Claude Code with your Pro or Max plan](https://support.claude.com/en/articles/11145838-use-claude-code-with-your-pro-or-max-plan) — checked 2026-10-21
9. [Configure Claude Code permissions](https://code.claude.com/docs/en/permissions) — checked 2026-10-21
10. [Anthropic announces Series H financing](https://www.anthropic.com/news/series-h) — checked 2026-10-21
11. [Anthropic introduces Claude Code](https://www.anthropic.com/news/claude-3-7-sonnet) — checked 2026-10-21

---

# Claude Duet

> A proactive agent for ongoing professional work, daily briefings, and tasks that continue while you are away.

- **Maker:** Anthropic
- **Catalogue type:** Hosted agent
- **Website:** https://claude.com/product/claude-duet
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 8 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/claude-duet/
- **This record:** https://agentlist.wiki/agents/claude-duet.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Waitlist · some regions, paid plan | documented | 2 |
| Price & usage costs | Claude Pro $20/mo | documented | 1, 3 |
| Included in another plan | Claude Pro / Max | limited | 1 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | Cloud | documented | 2 |
| Works while your computer is off | Keeps working | documented | 2 |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 7, 8 |
| Company type | Major AI lab | documented | 6 |
| Launch date | Oct 21, 2026 | documented | 2 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Major AI lab [6]
- **Company stage** `documented` — Series H [6]
- **Disclosed funding** `documented` — US$65B · Series H · May 28, 2026 [6]
- **Launch date** `documented` — Oct 21, 2026 · Research preview [2]

**Launch history**

- 2026-10-21 · Research preview [2]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Morning work briefings; carries over Claude memory, skills, and connected apps. [1]
- **Computer use** `documented` — Own cloud computer; does not require your device to remain on. [2]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `documented` — Its own cloud computer. Whether you can watch or take over its screen is not specified. [2]
- **Works while your computer is off** `documented` — Runs on its own cloud computer and does not require your device to remain on. [2]
- **Browser control** `limited` (partial or conditional only) — Uses its own cloud browser; viewing or taking over that browser is not yet available. [2]
- **Voice & ambient listening** `documented` — Voice messages in the ongoing conversation; ambient listening is not specified. [1]
- **Background work & triggers** `documented` — Daily briefs and background checks [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Desktop web at claude.ai; no native Duet desktop app is listed. [1]
- **iPhone & Android apps** `documented` — Standalone Duet apps for iOS and Android. [1]
- **Web & messaging** `documented` — Web, iOS, Android; text and voice [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Claude Opus 5.5 [1]
- **Bring your own model** `documented` — Opus 5.5 is included through the Claude subscription; no BYO-model option is listed. [1]
- **Included in another plan** `limited` (partial or conditional only) — Included in Claude Pro and Max during the US research preview; shares Claude usage limits. [1]
- **Price & usage costs** `documented` — US research preview included in Claude Pro (US$20/month or US$200/year) and Max (US$100 or US$200/month). Duet's background work shares Claude plan usage. [1, 3]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Settings, then Privacy, then Export data, on Free, Pro and Max plans. Deleted data is removed from back-end storage within 30 days. [7, 8]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed cloud computer [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Credential isolation is built in. 1Password saved-login integration is announced as coming soon. [2]
- **Approval before actions** `documented` — Waits before sending, paying, signing, or deleting; some financial actions remain unavailable. [2]
- **What it remembers** `documented` — Claude account memory, skills, and one ongoing thread [1]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — Approval before sensitive actions; visible activity [1]
- **Availability & access** `documented` — US Pro / Max waitlist research preview [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Claude Pro | US$20/month | Monthly, or US$200/year | Duet research preview in the US; Claude subscription and Claude Code | Waitlist applies; Duet tasks and background checks share Claude usage limits | 1, 3 |
| Claude Max 5x | US$100/month | Monthly web subscription | Duet preview; five times Pro's session allowance; US$100/month in claimable Claude Platform API credits | Claude session and weekly limits apply; API credits require seven days on the plan; do not extend Duet usage | 1, 4, 5 |
| Claude Max 20x | US$200/month | Monthly web subscription | Duet preview; twenty times Pro's session allowance; US$200/month in claimable Claude Platform API credits | Claude session and weekly limits apply; API credits require seven days on the plan; do not extend Duet usage | 1, 4, 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Desktop web at claude.ai; no native Duet desktop app is listed. | 1 |
| iOS | Available | Standalone Duet apps for iOS and Android. | 1 |
| Android | Available | Standalone Duet apps for iOS and Android. | 1 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-21 · Launch · [Claude Duet enters research preview](https://claude.com/resources/articles/claude-duet) · Anthropic — Anthropic opens a waitlist for a proactive personal agent, starting with US Pro and Max subscribers.

## Sources

1. [Claude Duet preview access and shared usage](https://claude.com/product/claude-duet) — checked 2026-10-21
2. [Introducing Claude Duet](https://claude.com/resources/articles/claude-duet) — checked 2026-10-21
3. [Choose a Claude plan](https://support.claude.com/en/articles/11049762-choose-a-claude-plan) — checked 2026-10-21
4. [Claude Max pricing and usage limits](https://support.claude.com/en/articles/11049741-what-is-the-max-plan) — checked 2026-10-21
5. [Monthly API credits bundled with Claude Max](https://support.claude.com/en/articles/17154008-monthly-api-credits-for-max-and-team-plans) — checked 2026-10-21
6. [Anthropic announces Series H financing](https://www.anthropic.com/news/series-h) — checked 2026-10-21
7. [Export your Claude data](https://privacy.claude.com/en/articles/9450526-export-your-claude-data) — checked 2026-10-22
8. [How long Anthropic stores personal data](https://privacy.claude.com/en/articles/10023548-how-long-do-you-store-personal-data) — checked 2026-10-22

---

# Codex

> OpenAI’s coding agent across desktop, terminal, IDE, cloud, and iOS, with local computer use, scheduled tasks, and optional local models.

- **Maker:** OpenAI
- **Catalogue type:** Coding tool (adjacent)
- **Website:** https://learn.chatgpt.com/docs/codex/cli
- **Source code:** https://github.com/openai/codex
- **Also known as:** unknown — no other names recorded
- **Evidence:** 19 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/codex/
- **This record:** https://agentlist.wiki/agents/codex.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · some regions, paid plan | limited | 10, 12, 13 |
| Price & usage costs | From $20/mo plan; API option | documented | 8, 14 |
| Included in another plan | ChatGPT Plus / Pro / business plans | documented | 8, 14 |
| Where it runs | Local or managed cloud | documented | 3, 7, 13 |
| Open source & license | Open-source CLI; hosted app separate | limited | 1 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | unknown | unknown | — |
| Company type | Major AI lab | documented | 18 |
| Launch date | Apr 16, 2025 | documented | 19 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Major AI lab [18]
- **Company stage** `documented` — Private [18]
- **Disclosed funding** `documented` — US$122B · Funding round · Mar 31, 2026 · Committed capital [18]
- **Launch date** `documented` — Apr 16, 2025 · Open-source release · Codex CLI [19]

**Launch history**

- 2025-04-16 · Open-source release · Codex CLI [19]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Local code work and cloud delegation, with an open CLI that supports custom or local model providers. [1, 3, 7]
- **Computer use** `limited` (partial or conditional only) — Local GUI control on macOS and Windows through the desktop app. Windows tasks use the foreground; Linux computer use is not available in the preview. [5, 10]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Built-in desktop browser plus browser-extension control of existing tabs; built-in browser is not available in CLI or IDE. [11]
- **Voice & ambient listening** `documented` — Desktop voice dictation turns speech into an editable prompt; ambient listening is not specified. [15]
- **Background work & triggers** `limited` (partial or conditional only) — Desktop recurring tasks; local project tasks need the computer on and app running. [17]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native ChatGPT app with Codex on macOS, Windows, and Linux preview. [5, 12]
- **iPhone & Android apps** `limited` (partial or conditional only) — Codex in ChatGPT for iOS can run tasks in cloud environments or connected computers. Android Codex support is not specified in the current mobile guide. [13]
- **Web & messaging** `documented` — Desktop, terminal, IDE, web cloud tasks, and iOS. [3, 12, 13]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Codex SDK and app-server interface provide programmatic local-agent control. [6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — OpenAI models, custom providers, or local models via Ollama and LM Studio. [7]
- **Bring your own model** `documented` — ChatGPT allowance or API credentials; configurable providers and local Ollama / LM Studio models. [7, 8]
- **Included in another plan** `documented` — Included usage with ChatGPT Plus, Pro, and eligible business plans; API-key usage billed separately. [8, 14]
- **Price & usage costs** `documented` — ChatGPT Plus starts at $20/month with Codex usage; Pro and business plans provide other allowances. API-key use is metered separately. [8, 14]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `limited` (partial or conditional only) — CLI is Apache-2.0 open source. That license does not describe the hosted product or desktop app. [1]
- **Where it runs** `documented` — Local desktop / CLI / IDE work or managed cloud tasks. The open CLI can use self-hosted model endpoints; hosted Codex is a separate service. [3, 7, 13]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Configurable sandbox and action approvals, with optional automatic review. [9]
- **What it remembers** `documented` — Optional local memory across chats, stored on the connected machine; separate from ChatGPT account memory. [16]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `limited` (partial or conditional only) — Public desktop, CLI, IDE, cloud, and iOS surfaces; specific tools depend on region, plan, and workspace. [10, 12, 13]

## Plans

unknown — no published plan or price list is recorded. See the Price & usage costs claim above.

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Desktop, terminal, IDE, web cloud tasks, and iOS. | 3, 12, 13 |
| macOS | Available | Native ChatGPT app with Codex on macOS, Windows, and Linux preview. | 5, 12 |
| Windows | Available | Native ChatGPT app with Codex on macOS, Windows, and Linux preview. | 5, 12 |
| Linux | Preview | Native ChatGPT app with Codex on macOS, Windows, and Linux preview. | 5, 12 |
| iOS | Available | Codex in ChatGPT for iOS can run tasks in cloud environments or connected computers. Android Codex support is not specified in the current mobile guide. | 13 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Codex official repository](https://github.com/openai/codex) — checked 2026-10-21
2. [ChatGPT and Codex glossary](https://learn.chatgpt.com/docs/glossary) — checked 2026-10-21
3. [Codex CLI](https://learn.chatgpt.com/docs/codex/cli) — checked 2026-10-21
4. [Remote connections](https://learn.chatgpt.com/docs/remote-connections) — checked 2026-10-21
5. [ChatGPT and Codex updates](https://learn.chatgpt.com/docs/whats-new) — checked 2026-10-21
6. [Codex SDK](https://learn.chatgpt.com/docs/codex-sdk) — checked 2026-10-21
7. [Custom and local model providers](https://learn.chatgpt.com/docs/config-file/config-advanced) — checked 2026-10-21
8. [Codex authentication](https://learn.chatgpt.com/docs/auth) — checked 2026-10-21
9. [Agent approvals and security](https://learn.chatgpt.com/docs/agent-approvals-security) — checked 2026-10-21
10. [Computer Use in ChatGPT Work and Codex](https://learn.chatgpt.com/docs/computer-use) — checked 2026-10-21
11. [Browser in ChatGPT Work and Codex](https://learn.chatgpt.com/docs/browser) — checked 2026-10-21
12. [ChatGPT desktop app](https://learn.chatgpt.com/docs/app) — checked 2026-10-21
13. [Codex on mobile](https://learn.chatgpt.com/docs/mobile) — checked 2026-10-21
14. [ChatGPT and Codex pricing](https://learn.chatgpt.com/docs/pricing) — checked 2026-10-21
15. [Voice dictation in the desktop app](https://learn.chatgpt.com/docs/prompting) — checked 2026-10-21
16. [Local Codex memories](https://learn.chatgpt.com/docs/customization/memories) — checked 2026-10-21
17. [Scheduled tasks](https://learn.chatgpt.com/docs/automations) — checked 2026-10-21
18. [OpenAI closes its March 2026 funding round](https://openai.com/index/accelerating-the-next-phase-ai/) — checked 2026-10-21
19. [OpenAI introduces Codex CLI](https://openai.com/index/introducing-o3-and-o4-mini/) — checked 2026-10-21

---

# Cue

> Agents with their own phone, email, wallet, and computer, continuing work after logout. This record refers specifically to cue.im.

- **Maker:** Manus AI
- **Catalogue type:** Hosted agent
- **Website:** https://www.cue.im/
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 1 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/cue/
- **This record:** https://agentlist.wiki/agents/cue.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Access unverified | limited | 1 |
| Price & usage costs | unknown | unknown | — |
| Included in another plan | unknown | unknown | — |
| Where it runs | Unverified | limited | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | unknown | unknown | — |
| Company type | unknown | unknown | — |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Each agent gets an email, phone, wallet and computer; multiple agents can share a group chat [1]
- **Computer use** `limited` (partial or conditional only) — Dedicated agent computer workspace; local versus cloud architecture is not specified [1]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Agents make and take calls from their own phone numbers [1]
- **Background work & triggers** `documented` — Connect services and define repeatable Routines; agents keep working after logout [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — macOS and Windows download options on the official site [1]
- **iPhone & Android apps** `limited` (partial or conditional only) — Mobile download offered; iPhone/Android availability not specified [1]
- **Web & messaging** `documented` — Agent workspace, group chat, email and phone [1]
- **Messaging services** `documented` — Email [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `limited` (partial or conditional only) — Dedicated agent workspace; underlying hosting unverified [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — You choose access permissions and which actions need review [1]
- **What it remembers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — User-defined access and approval boundaries [1]
- **Availability & access** `limited` (partial or conditional only) — Official product site; access requirements unverified [1]

## Plans

unknown — no published plan or price list is recorded. See the Price & usage costs claim above.

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | macOS and Windows download options on the official site | 1 |
| Windows | Available | macOS and Windows download options on the official site | 1 |
| Mobile app | Limited availability | Mobile download offered; iPhone/Android availability not specified | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Email | Available | Agent workspace, group chat, email and phone | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Cue personal agents](https://www.cue.im/) — checked 2026-10-21

---

# Eney

> A Mac and iPhone assistant that handles app and file tasks, captures meeting notes, and suggests what to do next.

- **Maker:** MacPaw
- **Catalogue type:** Hosted agent
- **Website:** https://eney.ai/
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 5 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/eney/
- **This record:** https://agentlist.wiki/agents/eney.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Trial, then $14.99/mo | documented | 1 |
| Included in another plan | Setapp AI credits | documented | 2 |
| Where it runs | Local + cloud models | documented | 3 |
| Open source & license | Proprietary | documented | 4 |
| Controls your computer | Yes · Mac | documented | 3 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | Partly | limited | 3 |
| If the vendor stops | Partly | limited | 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 3 |
| Company type | Established company | documented | 2 |
| Launch date | May 22, 2025 | documented | 5 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Established company [2]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — May 22, 2025 · Announcement · Original Mac announcement [5]

**Launch history**

- 2025-05-22 · Announcement · Original Mac announcement [5]
- 2026-10-15 · Public launch · Mac and iPhone release [2]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Mac-aware suggestions, screen selection with Eney Wand, meeting notes and ready-made workflows. [2]
- **Computer use** `documented` — Local Mac tools read files, run commands and act across connected apps. [3]
- **Controls your computer** `documented` — Local tools on the Mac read files, run commands and act across connected apps. [3]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `limited` (partial or conditional only) — Mac actions execute locally and need the Mac running; signed-in chats sync and model requests run in the cloud. [3]
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Voice input, read-aloud replies and meeting transcription. [2, 3]
- **Background work & triggers** `documented` — Workflows run on demand or on a schedule, including morning briefs and weekly reviews. [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native Mac app; Apple silicon and macOS 15 or later. [1]
- **iPhone & Android apps** `documented` — Native iPhone app. [1]
- **Web & messaging** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — OpenAI for Fast mode; Claude for More thorough and iPhone; Gemini for image generation. [3]
- **Bring your own model** `documented` — Cloud models are supplied through the Setapp AI gateway; no local inference. [3]
- **Included in another plan** `documented` — Uses Setapp AI credits; a Setapp app-library membership is not required. [2]
- **Price & usage costs** `documented` — 14-day trial with 10,000 credits, then $14.99–$49.99/month. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Mac actions run locally but model requests go to the cloud, so the local half may outlive the service and the rest would not. [3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Memories can be exported to a file and imported again. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — Proprietary software under a limited, non-transferable license. [4]
- **Where it runs** `documented` — Mac actions execute locally; model requests run in the cloud and signed-in chats sync. [3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Asks before writes by default; permitted read-only tools run without repeated approval. [3]
- **What it remembers** `documented` — Mnemos stores memories locally and updates corrected or outdated facts. [2, 3]
- **Edit & export your data** `documented` — Memories can be imported from a file and exported. [3]
- **Privacy & isolation** `documented` — Permission-based app/file access; provider training disabled; synced chats are not end-to-end encrypted. [3]
- **Availability & access** `documented` — Available for Mac and iPhone; released October 15, 2026. [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Trial | Free | One trial | 10,000 AI credits | Ends after 14 days or when credits run out | 4 |
| Starter | $14.99/month | USD, billed monthly; tax included | 10,000 AI credits per month | Unused monthly credits do not roll over | 1 |
| Boost | $29.99/month | USD, billed monthly; tax included | 25,000 AI credits per month | Unused monthly credits do not roll over | 1 |
| Power | $49.99/month | USD, billed monthly; tax included | 50,000 AI credits per month | Unused monthly credits do not roll over | 1 |
| Credit top-up | Shown at checkout | One-time purchase | Additional Setapp AI credits | 12-month credit package | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Apple silicon; macOS 15 or later. | 1 |
| iOS | Available | Available on the iPhone App Store. | 1 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` Not established: No dedicated hardware is specified in the reviewed product pages.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-15 · Launch · [Eney launches for Mac and iPhone](https://macpaw.com/news/eney-announcement) — MacPaw releases Eney with proactive workflows, Mac tools and a 14-day credit trial.

## Sources

1. [Eney features, platforms and credit plans](https://eney.ai/) — checked 2026-10-21
2. [MacPaw launches Eney for Mac and iPhone](https://macpaw.com/news/eney-announcement) — checked 2026-10-21
3. [Eney privacy notice](https://macpaw.com/legal/eney-privacy-policy) — checked 2026-10-21
4. [Eney terms and credit rules](https://macpaw.com/legal/eney-terms-of-service) — checked 2026-10-21
5. [MacPaw introduces Eney in May 2025](https://macpaw.com/news/why-we-built-eney) — checked 2026-10-21

---

# Gemini Tasks

> A cloud personal agent connected to Workspace, designed to work in the background under user direction.

- **Maker:** Google
- **Catalogue type:** Hosted agent
- **Website:** https://blog.google/products-and-platforms/products/workspace/workspace-updates/
- **Source code:** unknown — no public repository recorded
- **Also known as:** Gemini Spark, Spark, Tasks in Gemini
- **Evidence:** 17 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/gemini-spark/
- **This record:** https://agentlist.wiki/agents/gemini-spark.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · paid plan, age limits | documented | 8 |
| Price & usage costs | AI Pro $19.99/mo | documented | 8, 9, 10 |
| Included in another plan | Google AI Pro / Ultra | documented | 8 |
| Where it runs | Google Cloud VMs | documented | 2 |
| Open source & license | unknown | unknown | — |
| Controls your computer | Limited | limited | 4 |
| Its own computer | Cloud | documented | 2 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 2 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 16, 17 |
| Company type | Public company | documented | 11 |
| Launch date | May 19, 2026 | documented | 12 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Public company · part of Alphabet [11]
- **Company stage** `documented` — Public · NASDAQ: GOOGL / GOOG [11]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — May 19, 2026 · Announcement · Introduced as Gemini Spark [12]

**Launch history**

- 2026-05-19 · Announcement · Introduced as Gemini Spark [12]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Deep Gmail, Docs, and Sheets integration; trigger workflows from incoming email. [3]
- **Computer use** `limited` (partial or conditional only) — Cloud agent; desktop control through the macOS app was announced, but completion of that rollout is not established by the reviewed sources. [4]
- **Controls your computer** `limited` (partial or conditional only) — Desktop control through the macOS app was announced, but completion of that rollout is not established by the sources reviewed. [4]
- **Its own computer** `documented` — Dedicated Google Cloud virtual machines. Whether you can watch or drive one is not specified. [2]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Chrome auto browse works with signed-in accounts; the feature initially rolled out in the US. [5]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Tasks can run on a schedule or an event trigger, with up to 15 tasks executing concurrently. A scheduled run does not start if all 15 slots are occupied. [8]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Gemini has macOS and Windows apps; Tasks support on the desktop was announced as a later rollout. Web access is available. [4, 15]
- **iPhone & Android apps** `limited` (partial or conditional only) — Task management in the Gemini apps for iOS and Android; account and feature availability apply. [6, 7]
- **Web & messaging** `documented` — Gemini app with Workspace integrations [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Gemini 3.6 Flash in July rollout [3]
- **Bring your own model** `documented` — Gemini model usage included through Google AI plans; BYO model keys are not listed for Spark. [3]
- **Included in another plan** `documented` — Included with Google AI Pro or Ultra on eligible personal Google accounts. [8]
- **Price & usage costs** `documented` — Google AI Pro is US$19.99/month in the US. Tasks are also included with Google AI Ultra; Ultra's current price varies by tier and market. [8, 9, 10]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [2]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Export through Google Takeout. Deletion generally takes around two months; temporary chats and those recorded with Keep Activity off are held 72 hours, and reviewed chats up to three years. [16, 17]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Dedicated Google Cloud virtual machines [2]
- **Passwords & credentials** `limited` (partial or conditional only) — Uses saved passwords through Chrome with permission; no separate Spark vault is described. [5]
- **Approval before actions** `documented` — Asks before sensitive actions such as spending money or sending email; Chrome hands sensitive steps back to you. [3, 5]
- **What it remembers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — User-controlled connections; approval for high-stakes actions [3]
- **Availability & access** `documented` — Available to adults with eligible personal Google accounts, Google AI Pro or Ultra, Keep Activity enabled, and Personal Intelligence enabled. Excludes EEA, Nigeria, Switzerland and the UK. [8]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Google AI Pro | US$19.99/month | Monthly US price | Tasks in Gemini; 5TB storage; Gemini 4 Argon, Google app features and Deep Research; YouTube Premium Lite and Google Home Premium Standard where available | Up to 15 simultaneous tasks; Compute allowance shared across Gemini tools | 8, 9, 10 |
| Google AI Ultra | Not published | Monthly; regional price and tier shown at signup | Tasks in Gemini with higher compute limits; Ultra 5×: 20TB storage; Ultra 20×: 30TB; YouTube Premium where available | Up to 15 simultaneous tasks; Tier multipliers compare overall Gemini usage to Pro, not task counts | 8, 9 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Gemini has a macOS app; Spark support was announced as a later rollout. Web access is available. | 4 |
| macOS | Coming soon | Gemini has a macOS app; Spark support was announced as a later rollout. Web access is available. | 4 |
| iOS | Limited availability | Task management in the Gemini apps for iOS and Android; account and feature availability apply. | 6, 7 |
| Android | Limited availability | Task management in the Gemini apps for iOS and Android; account and feature availability apply. | 6, 7 |
| Windows | Available | Gemini app for Windows, announced 10 September 2026. | 15 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

| Data source | Open to third parties? | This agent's route | Detail | Sources |
| --- | --- | --- | --- | --- |
| Google Photos library | First-party only | First-party | Reads the Google Photos library directly. Google removed the equivalent third-party scopes after March 2025, so no other agent can follow. | 13, 14 |
| Gmail | Open, with review | First-party | Reads Gmail natively. Third parties can reach the same mail, but only through restricted scopes with Google verification and an annual security assessment. | 13 |

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 3981 posts, 457 positive and 449 negative rated user-weeks — 50.4% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Getting tasks done | 219 | 112 | 66.2% | https://feedbackbench.com/criteria/act.md |
| Connecting accounts, sites and devices | 59 | 89 | 39.9% | https://feedbackbench.com/criteria/connect.md |
| Getting access and set up | 49 | 73 | 40.2% | https://feedbackbench.com/criteria/access.md |
| Running on its own | 64 | 31 | 67.4% | https://feedbackbench.com/criteria/auto.md |
| Paying for the agent | 31 | 44 | 41.3% | https://feedbackbench.com/criteria/billing.md |
| Talking to the agent | 23 | 47 | 32.9% | https://feedbackbench.com/criteria/ask.md |
| Staying in control | 29 | 41 | 41.4% | https://feedbackbench.com/criteria/control.md |
| Dealing with the vendor | 8 | 52 | 13.3% | https://feedbackbench.com/criteria/vendor.md |
| Trusting it with data and safety | 3 | 24 | 11.1% | https://feedbackbench.com/criteria/trust.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-10 · Release · [The Gemini app comes to Windows](https://blog.google/innovation-and-ai/products/gemini-app/gemini-app-now-on-windows/) · Google — Google ships a Gemini desktop app for Windows, alongside the existing macOS app.
- 2026-07-29 · Availability change · [Gemini Spark expands to India](https://blog.google/intl/en-in/company-news/technology/introducing-gemini-spark-your-247-personal-ai-agent-in-country/) · Google — Google announces a rollout for AI Pro subscribers and describes scheduled and event-driven Workspace tasks.

## Ecosystem

- Built on: Antigravity → Gemini Tasks — Google introduced Gemini Spark on the Antigravity harness. The product is now called Gemini Tasks. (https://blog.google/innovation-and-ai/technology/ai/google-io-2026-all-our-announcements/)

## Sources

1. [Workspace updates · Google](https://blog.google/products-and-platforms/products/workspace/workspace-updates/) — checked 2026-10-21
2. [Google I/O 2026 keynote · Google](https://blog.google/innovation-and-ai/sundar-pichai-io-2026/) — checked 2026-10-21
3. [Gemini Spark in India · Google](https://blog.google/intl/en-in/company-news/technology/introducing-gemini-spark-your-247-personal-ai-agent-in-country/) — checked 2026-10-21
4. [The Gemini app becomes more agentic](https://blog.google/innovation-and-ai/products/gemini-app/next-evolution-gemini-app/) — checked 2026-10-21
5. [Gemini Spark now integrates with Chrome](https://blog.google/innovation-and-ai/products/gemini-app/gemini-spark-updates-july-2026/) — checked 2026-10-21
6. [Manage Gemini tasks on iPhone and iPad](https://support.google.com/gemini/answer/17094196?co=GENIE.Platform%3DiOS&hl=en) — checked 2026-10-21
7. [Manage Gemini tasks on Android](https://support.google.com/gemini/answer/17094196?co=GENIE.Platform%3DAndroid&hl=en) — checked 2026-10-21
8. [Tasks in Gemini: Spark rename, eligibility and limits](https://support.google.com/gemini/answer/17094507) — checked 2026-10-21
9. [Google AI plan benefits and Ultra tiers](https://one.google.com/about/google-ai-plans/) — checked 2026-10-21
10. [Google One US monthly pricing](https://one.google.com/about/plans) — checked 2026-10-21
11. [Alphabet annual report: listed shares and Google ownership](https://abc.xyz/assets/99/21/46cafdba41089a12a2d86ea47d44/goog026-annualreport2024-web.pdf) — checked 2026-10-21
12. [Google I/O announcement: Gemini Spark and Antigravity](https://blog.google/innovation-and-ai/technology/ai/google-io-2026-all-our-announcements/) — checked 2026-10-21
13. [Gemini Personal Intelligence: Gmail, Photos, YouTube and Search](https://blog.google/innovation-and-ai/products/gemini-app/personal-intelligence/) — checked 2026-10-22
14. [Google Photos APIs: library scopes removed](https://developers.google.com/photos/support/updates) — checked 2026-10-22
15. [The Gemini app is now on Windows](https://blog.google/innovation-and-ai/products/gemini-app/gemini-app-now-on-windows/) — checked 2026-10-22
16. [Gemini Apps data controls and export](https://support.google.com/gemini/answer/13594961) — checked 2026-10-22
17. [Google data retention](https://policies.google.com/technologies/retention) — checked 2026-10-22

---

# Grok Bot

> Named AI teammates that work across apps on a persistent computer and collaborate on ongoing projects.

- **Maker:** Grok / Cursor
- **Catalogue type:** Hosted agent
- **Website:** https://x.ai/bot
- **Source code:** unknown — no public repository recorded
- **Also known as:** Grokbot
- **Evidence:** 14 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/grok-bot/
- **This record:** https://agentlist.wiki/agents/grok-bot.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | From US$30/mo | documented | 8, 9 |
| Included in another plan | X Premium / Super / Ultra; legacy plans | documented | 8 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | Cloud | documented | 3 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | None described | documented | 14 |
| Company type | Major AI lab | documented | 11 |
| Launch date | Aug 11, 2026 | documented | 11 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Major AI lab [11]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Aug 11, 2026 · Public launch [11]

**Launch history**

- 2026-08-11 · Public launch [11]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Teach a workflow by demonstrating it; several named Bots can coordinate on shared work. [1]
- **Computer use** `documented` — Persistent cloud desktop; each Bot has a screen on the shared user computer. [3]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `documented` — A persistent cloud desktop, with a screen for each Bot on a computer shared across your Bots. [3]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Cloud browser with persistent logins; browser sessions are shared among your Bots. [3]
- **Voice & ambient listening** `documented` — Dictation, voice chat, and voice memos. [2]
- **Background work & triggers** `documented` — Scheduled routines and 24/7 work [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native Grok Bot applications for macOS, Windows, and Linux. [4]
- **iPhone & Android apps** `documented` — Companion apps for iOS and Android. [5]
- **Web & messaging** `documented` — Desktop and mobile; text and voice [2]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — Admin API manages team access, policy, and computers; a public API for chatting with your personal Bot is not specified. [6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `documented` — Weekly model usage is included with eligible plans; bringing your own model key is not specified. [2]
- **Included in another plan** `documented` — Bundled with The X Subscription Premium, Super and Ultra; eligible legacy Cursor and linked SuperGrok plans also grant access. [8]
- **Price & usage costs** `documented` — New customers: The X Subscription Premium US$30/month, Super US$100/month or Ultra US$200/month; annual discounts available. Eligible legacy Cursor and linked SuperGrok plans continue. [8, 9]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — The privacy policy describes rights to access, correct, update and delete, but no export or portability right. Account deletion removes data within 30 days. [14]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed cloud; bots share a computer per user [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Supports masked secret requests and manual sign-in takeover; all Bots share signed-in sessions. A personal password-manager vault is not specified. [3]
- **Approval before actions** `documented` — Ask-first rules and automatic review; allow once, save a rule, or deny an action. [7]
- **What it remembers** `documented` — Context accumulates across work [2]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — Action approvals; isolation is per user, not per bot [1]
- **Availability & access** `documented` — Individual, team, and enterprise access [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| X Premium | US$30/month | Monthly, or US$300/year | Grok Bot, Grok, Cursor and linked X benefits; 4× base included usage | Weekly allowance shared across products; Extra usage uses prepaid credits | 8, 9 |
| X Super | US$100/month | Monthly, or US$1,000/year | Premium features; 15× base included usage | Shared weekly allowance; paid top-ups | 9 |
| X Ultra | US$200/month | Monthly, or US$2,000/year | Premium features; 50× base included usage | Shared weekly allowance; paid top-ups | 9 |
| Legacy Cursor Pro / Pro+ / Ultra | US$20 / US$60 / US$200 per month | Existing Cursor plans | Grok Bot with progressively larger weekly allowances | Grok links and Cursor allowances do not stack; Optional overage billed through Cursor | 8, 10 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Native Grok Bot applications for macOS, Windows, and Linux. | 4 |
| Windows | Available | Native Grok Bot applications for macOS, Windows, and Linux. | 4 |
| Linux | Available | Native Grok Bot applications for macOS, Windows, and Linux. | 4 |
| iOS | Available | Companion apps for iOS and Android. | 5 |
| Android | Available | Companion apps for iOS and Android. | 5 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

| Data source | Open to third parties? | This agent's route | Detail | Sources |
| --- | --- | --- | --- | --- |
| X public posts | Open to third parties | First-party | Reads X posts as part of xAI. This is not exclusive: xAI sells an X Search tool and X sells an enterprise firehose of all public posts, and nothing documents Grok receiving data its API customers cannot buy. | 12, 13 |

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 55642 posts, 9150 positive and 6348 negative rated user-weeks — 59% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: App Store, Play Store, Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Getting tasks done | 3270 | 921 | 78% | https://feedbackbench.com/criteria/act.md |
| Paying for the agent | 850 | 2980 | 22.2% | https://feedbackbench.com/criteria/billing.md |
| Running on its own | 2051 | 723 | 73.9% | https://feedbackbench.com/criteria/auto.md |
| Talking to the agent | 766 | 778 | 49.6% | https://feedbackbench.com/criteria/ask.md |
| Connecting accounts, sites and devices | 596 | 779 | 43.3% | https://feedbackbench.com/criteria/connect.md |
| Getting access and set up | 453 | 655 | 40.9% | https://feedbackbench.com/criteria/access.md |
| Staying in control | 481 | 551 | 46.6% | https://feedbackbench.com/criteria/control.md |
| Dealing with the vendor | 200 | 715 | 21.9% | https://feedbackbench.com/criteria/vendor.md |
| Trusting it with data and safety | 97 | 253 | 27.7% | https://feedbackbench.com/criteria/trust.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-08-26 · Availability change · [Grok Bot expands to more subscription plans](https://x.ai/news/grok-bot-more-plans) · Grok — Access expands to SuperGrok, Cursor Pro, and Cursor Teams plans, with a separate Bot usage allowance.

## Sources

1. [Grok Bot · Official product page](https://x.ai/bot) — checked 2026-10-21
2. [Grok Bot overview · Official docs](https://docs.x.ai/grok-bot/overview) — checked 2026-10-21
3. [Use the computer and apps](https://docs.x.ai/grok-bot/computer-and-apps) — checked 2026-10-21
4. [Get started with Grok Bot](https://docs.x.ai/grok-bot/get-started) — checked 2026-10-21
5. [Grok Bot for Mobile](https://docs.x.ai/grok-bot/mobile) — checked 2026-10-21
6. [Grok Bot for teams and enterprises](https://docs.x.ai/grok-bot/teams-and-enterprises) — checked 2026-10-21
7. [Approvals, security, and privacy](https://docs.x.ai/grok-bot/approvals-security-and-privacy) — checked 2026-10-21
8. [Grok Bot plans, linked subscriptions and weekly usage](https://cursor.com/help/grok-bot/plans) — checked 2026-10-21
9. [The X Subscription prices](https://docs.x.ai/x-subscription/pricing) — checked 2026-10-21
10. [Legacy Cursor plan prices](https://cursor.com/docs/account/pricing/legacy-cursor-plans) — checked 2026-10-21
11. [Grok Bot availability in the developer release notes](https://docs.x.ai/developers/release-notes) — checked 2026-10-21
12. [xAI X Search tool pricing](https://docs.x.ai/developers/tools/x-search) — checked 2026-10-22
13. [X Enterprise API: full public post firehose](https://docs.x.com/enterprise-api/introduction) — checked 2026-10-22
14. [xAI privacy policy](https://x.ai/legal/privacy-policy) — checked 2026-10-22

---

# Hark Pro

> A personal agent combining a continuing conversation, memory, and a cloud browser called Handoff.

- **Maker:** Hark Labs
- **Catalogue type:** Hosted agent
- **Website:** https://hark.com
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 7 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/hark/
- **This record:** https://agentlist.wiki/agents/hark.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Free; $20–$100/mo | documented | 2 |
| Included in another plan | Standalone Hark plans | documented | 2 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | Cloud | documented | 3 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 7 |
| Company type | Startup | documented | 5 |
| Launch date | Oct 6, 2026 | documented | 4, 6 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [5]
- **Company stage** `documented` — Series A [5]
- **Disclosed funding** `documented` — US$700M+ · Series A · May 2026 [5]
- **Launch date** `documented` — Oct 6, 2026 · Public launch [4, 6]

**Launch history**

- 2026-10-06 · Public launch [4, 6]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Action Buttons handle suggested chores in one tap; custom Panels are mini-apps built inside Hark. [1]
- **Computer use** `limited` (partial or conditional only) — Handoff runs on a personal cloud computer; published workflows focus on websites, not local desktop control. [3]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `documented` — A personal cloud computer runs Handoff tasks. Whether you can watch its screen is not specified. [3]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Full cloud browser can sign in, complete forms, and place orders. [1]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Scheduled tasks and reusable Specialties [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Web app on desktop; the product lists no native desktop app. [4]
- **iPhone & Android apps** `documented` — Native iOS and Android apps. [4]
- **Web & messaging** `documented` — Hark app and web [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `documented` — Service includes model usage; free and paid plans differ in usage allowance. BYO keys are not listed. [2]
- **Included in another plan** `documented` — Standalone Hark subscription: free tier and paid usage upgrades. [2]
- **Price & usage costs** `documented` — Free usage cap; $20 / $100 monthly tiers increase usage [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A portability right to information in a structured, commonly used, machine-readable format, exercised through privacy settings rather than a download. General retention lasts for the lifetime of the account. [7]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed cloud computer through Handoff [1]
- **Passwords & credentials** `documented` — Built-in Secured by Hark vault stores encrypted logins and payment details; values are supplied only to the destination page. [1]
- **Approval before actions** `documented` — Checks in before actions requiring your approval; user-facing Action Buttons can authorize suggested chores. [3]
- **What it remembers** `documented` — Conversation memory; ask it to forget [1]
- **Edit & export your data** `limited` (partial or conditional only) — Memory deletion documented; export not verified [1]
- **Privacy & isolation** `limited` (partial or conditional only) — Action Buttons and user-directed Handoffs; full control policy not assessed [1]
- **Availability & access** `documented` — Free and paid plans [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Hark Pro | US$0 | Free | Every product feature; no ads | Base usage allowance | 2 |
| Hark Pro² | US$20/month | Monthly | Same features as Free | Twice the base usage | 2 |
| Hark Pro³ | US$100/month | Monthly | Same features as Free | Ten times the base usage | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web app on desktop; the product lists no native desktop app. | 4 |
| iOS | Available | Native iOS and Android apps. | 4 |
| Android | Available | Native iOS and Android apps. | 4 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 5024 posts, 959 positive and 621 negative rated user-weeks — 60.7% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: App Store, Play Store, Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Getting tasks done | 487 | 147 | 76.8% | https://feedbackbench.com/criteria/act.md |
| Paying for the agent | 99 | 156 | 38.8% | https://feedbackbench.com/criteria/billing.md |
| Talking to the agent | 142 | 103 | 58% | https://feedbackbench.com/criteria/ask.md |
| Running on its own | 174 | 19 | 90.2% | https://feedbackbench.com/criteria/auto.md |
| Connecting accounts, sites and devices | 52 | 106 | 32.9% | https://feedbackbench.com/criteria/connect.md |
| Dealing with the vendor | 16 | 139 | 10.3% | https://feedbackbench.com/criteria/vendor.md |
| Getting access and set up | 33 | 69 | 32.4% | https://feedbackbench.com/criteria/access.md |
| Trusting it with data and safety | 18 | 67 | 21.2% | https://feedbackbench.com/criteria/trust.md |
| Staying in control | 50 | 34 | 59.5% | https://feedbackbench.com/criteria/control.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-06 · Launch · [Hark Pro launches on web, iOS and Android](https://hark.com/articles/introducing-hark-pro) · Hark Labs — Hark Labs introduces Hark Pro across web and both mobile platforms.

## Sources

1. [Using Hark Pro · Hark](https://hark.com/using-hark) — checked 2026-10-21
2. [Hark subscriptions](https://hark.com/pricing) — checked 2026-10-21
3. [Hark Pro product overview](https://hark.com/) — checked 2026-10-21
4. [Introducing Hark Pro](https://hark.com/articles/introducing-hark-pro) — checked 2026-10-21
5. [Hark announces Series A financing](https://hark.com/articles/hark-announces-usd700-million-fundraising-round) — checked 2026-10-21
6. [Hark Pro public release date](https://techcrunch.com/2026/10/06/hark-releases-an-ai-personal-assistant-with-a-focus-on-privacy/) — checked 2026-10-21
7. [Hark privacy policy](https://hark.com/privacy-policy) — checked 2026-10-22

---

# Hermes Agent

> A persistent assistant that retains knowledge and develops reusable skills through experience.

- **Maker:** Nous Research
- **Catalogue type:** Self-hosted runtime
- **Website:** https://hermes-agent.nousresearch.com
- **Source code:** https://github.com/NousResearch/hermes-agent
- **Also known as:** unknown — no other names recorded
- **Evidence:** 9 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/hermes/
- **This record:** https://agentlist.wiki/agents/hermes.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1 |
| Included in another plan | Optional Nous Portal model + tools subscription | documented | 1 |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT | documented | 1 |
| Controls your computer | Yes · Mac, Windows, Linux | documented | 3 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | Startup | documented | 9 |
| Launch date | Feb 2026 | documented | 9 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [9]
- **Company stage** `documented` — Venture-backed [9]
- **Disclosed funding** `documented` — US$90M · Funding round · Oct 2026 [9]
- **Launch date** `documented` — Feb 2026 · Open-source release [9]

**Launch history**

- 2026-02 · Open-source release [9]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Learns reusable skills from experience; imports OpenClaw settings, memory, and skills. [1]
- **Computer use** `documented` — Controls macOS, Windows, and Linux desktops; can target the host or a supported sandbox. [3]
- **Controls your computer** `documented` — Drives the desktop on macOS, Windows and Linux, targeting either the host machine or a supported sandbox. [3]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Local or cloud browser automation; can connect to an existing Chromium browser. [4]
- **Voice & ambient listening** `documented` — Microphone conversations, spoken replies, Discord voice channels, and optional wake-word activation. [8]
- **Background work & triggers** `documented` — Built-in cron with cross-platform delivery. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native macOS (Apple Silicon), Windows, and Linux apps. [6]
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — Telegram, Discord, Slack, WhatsApp, Signal, CLI. [1]
- **Messaging services** `documented` — WhatsApp, Telegram, Signal, Slack, Discord [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Self-hosted OpenAI-compatible HTTP API with streaming and persistent sessions. [7]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Nous Portal, OpenRouter, OpenAI, custom endpoints, and others. [1]
- **Bring your own model** `documented` — Bring your own provider keys or model endpoint; provider and tool credentials are configurable. [1]
- **Included in another plan** `documented` — Optional Nous Portal subscription includes models, web search, image generation, speech, and cloud browsing. [1]
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [1]
- **Where it runs** `documented` — Local, VPS, containers, or serverless backends. [1]
- **Passwords & credentials** `documented` — Built-in encrypted login vault, browser autofill, and 1Password/Bitwarden integration. [5]
- **Approval before actions** `documented` — Configurable dangerous-command approvals; every saved-card autofill asks first. [2, 5]
- **What it remembers** `documented` — Curated memory, session search, learned skills, and user modeling. [1]
- **Edit & export your data** `limited` (partial or conditional only) — Agent Skills standard; multiple execution backends. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — User authorization, command approvals, and configurable container isolation. [2]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT open-source license; Bring your own provider keys or model endpoint; provider and tool credentials are configurable. | Usage limits and running costs depend on your chosen provider and deployment. | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Native macOS (Apple Silicon), Windows, and Linux apps. | 6 |
| Windows | Available | Native macOS (Apple Silicon), Windows, and Linux apps. | 6 |
| Linux | Available | Native macOS (Apple Silicon), Windows, and Linux apps. | 6 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | Telegram, Discord, Slack, WhatsApp, Signal, CLI. | 1 |
| Telegram | Available | Telegram, Discord, Slack, WhatsApp, Signal, CLI. | 1 |
| Signal | Available | Telegram, Discord, Slack, WhatsApp, Signal, CLI. | 1 |
| Slack | Available | Telegram, Discord, Slack, WhatsApp, Signal, CLI. | 1 |
| Discord | Available | Telegram, Discord, Slack, WhatsApp, Signal, CLI. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

| Date | Thread | Platform | Kind | Sentiment | Summary |
| --- | --- | --- | --- | --- | --- |
| 2026-06-05 | [Hermes user experiences: homelabs, memory and interface friction](https://news.ycombinator.com/item?id=48419000) | Hacker News | community | mixed | Users report useful homelab and Matrix automation alongside slow startup and CLI complaints. The thread also flags its submitted website as unofficial. |

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 36804 posts, 5389 positive and 3094 negative rated user-weeks — 63.5% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Running on its own | 1557 | 467 | 76.9% | https://feedbackbench.com/criteria/auto.md |
| Getting tasks done | 1224 | 536 | 69.5% | https://feedbackbench.com/criteria/act.md |
| Staying in control | 1182 | 419 | 73.8% | https://feedbackbench.com/criteria/control.md |
| Talking to the agent | 551 | 735 | 42.8% | https://feedbackbench.com/criteria/ask.md |
| Paying for the agent | 546 | 573 | 48.8% | https://feedbackbench.com/criteria/billing.md |
| Getting access and set up | 302 | 620 | 32.8% | https://feedbackbench.com/criteria/access.md |
| Connecting accounts, sites and devices | 442 | 323 | 57.8% | https://feedbackbench.com/criteria/connect.md |
| Dealing with the vendor | 266 | 470 | 36.1% | https://feedbackbench.com/criteria/vendor.md |
| Trusting it with data and safety | 139 | 164 | 45.9% | https://feedbackbench.com/criteria/trust.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-21 · Release · [Hermes Agent v0.22.4](https://github.com/NousResearch/hermes-agent/releases/tag/v0.22.4) · Nous Research — A patch release across the CLI, installer, Microsoft Store app, Docker and Hermes Cloud, removing suggestion pills from the desktop composer and fixing plugin and dashboard issues.
- 2026-10-13 · Release · [Hermes v0.22.0 adds guided onboarding and local models](https://github.com/NousResearch/hermes-agent/releases/tag/v0.22.0) — Release notes introduce conversational desktop setup, connectors, and a managed llama.cpp runtime. Notes date the release October 13; GitHub publication is October 14 UTC.
- 2026-10-08 · Availability change · [Hermes Agent arrives in the Microsoft Store](https://x.com/NousResearch/status/2108231536772596193) · Nous Research — Nous Research makes Hermes Agent installable from the Microsoft Store, with more Windows work signalled.
- 2026-04-19 · Release · [OpenFang v0.6.0 credits Hermes-inspired features](https://github.com/RightNow-AI/openfang/releases/tag/v0.6.0) — The release adds multi-destination cron delivery, skill configuration templates, and a shared slash-command registry.

## Sources

1. [Hermes Agent README](https://raw.githubusercontent.com/NousResearch/hermes-agent/main/README.md) — checked 2026-10-21
2. [Hermes security model](https://hermes-agent.nousresearch.com/docs/user-guide/security) — checked 2026-10-21
3. [Hermes computer use](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/features/computer-use.md) — checked 2026-10-21
4. [Hermes browser automation](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/features/browser.md) — checked 2026-10-21
5. [Hermes passwords and logins](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/features/credential-vault.md) — checked 2026-10-21
6. [Hermes Desktop](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/desktop.md) — checked 2026-10-21
7. [Hermes API server](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/features/api-server.md) — checked 2026-10-21
8. [Hermes voice mode](https://github.com/NousResearch/hermes-agent/blob/main/website/docs/user-guide/features/voice-mode.md) — checked 2026-10-21
9. [Nous Research: October funding and Hermes release history](https://nousresearch.com/a-note-on-our-fundraise) — checked 2026-10-21

---

# Instinct

> A personal assistant for errands and bookings through text or calls. Durable memory and background scheduling need further documentation.

- **Maker:** Spear Street Technology, Inc.
- **Catalogue type:** Under review
- **Website:** https://instinct.com
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 3 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/instinct/
- **This record:** https://agentlist.wiki/agents/instinct.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Access unverified | limited | 1 |
| Price & usage costs | unknown | unknown | — |
| Included in another plan | unknown | unknown | — |
| Where it runs | Unverified | limited | 2 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | None described | documented | 2 |
| Company type | Startup | documented | 3 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [3]
- **Company stage** `documented` — Series B · Aug 2026 [3]
- **Disclosed funding** `documented` — US$250M · Series B · Aug 26, 2026 · Total raised US$350M at a US$2.5B valuation; the round itself was US$250M. [3]
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Handles real-world errands such as grocery orders, travel bookings and sending flowers [1]
- **Computer use** `limited` (partial or conditional only) — Agent can operate a phone and computer; local versus cloud execution is not specified [1]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Phone conversations with the assistant [1]
- **Background work & triggers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Text and phone access; a native desktop app is not specified [1]
- **iPhone & Android apps** `limited` (partial or conditional only) — Text or call the assistant; a native mobile app is not specified [1]
- **Web & messaging** `documented` — Text and phone calls [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Neither the privacy policy nor the terms describe an export or portability right, and the terms state the company may but is not obliged to delete materials after account deletion. [2]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `limited` (partial or conditional only) — Managed service; execution architecture unverified [2]
- **Passwords & credentials** `limited` (partial or conditional only) — Can use third-party account credentials you supply; a password vault is not specified [2]
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `limited` (partial or conditional only) — Personalized context; durable memory unverified [2]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `limited` (partial or conditional only) — Connected-account access follows granted permissions; model-training opt-out is available [2]
- **Availability & access** `limited` (partial or conditional only) — Website links to app access; access requirements unverified [1]

## Plans

unknown — no published plan or price list is recorded. See the Price & usage costs claim above.

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

unknown — no native app or web client is recorded.

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 1535 posts, 198 positive and 72 negative rated user-weeks — 73.3% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Getting tasks done | 77 | 9 | 89.5% | https://feedbackbench.com/criteria/act.md |
| Talking to the agent | 32 | 11 | 74.4% | https://feedbackbench.com/criteria/ask.md |
| Running on its own | 28 | 7 | 80% | https://feedbackbench.com/criteria/auto.md |
| Trusting it with data and safety | 6 | 19 | 24% | https://feedbackbench.com/criteria/trust.md |
| Connecting accounts, sites and devices | 12 | 12 | 50% | https://feedbackbench.com/criteria/connect.md |
| Getting access and set up | 12 | 10 | 54.5% | https://feedbackbench.com/criteria/access.md |
| Dealing with the vendor | 2 | 15 | 11.8% | https://feedbackbench.com/criteria/vendor.md |
| Paying for the agent | 10 | 3 | 76.9% | https://feedbackbench.com/criteria/billing.md |
| Staying in control | 6 | 5 | 54.5% | https://feedbackbench.com/criteria/control.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Instinct](https://instinct.com/) — checked 2026-10-21
2. [Instinct privacy policy](https://instinct.com/privacy-policy) — checked 2026-10-21
3. [Instinct’s August 2026 Series B financing](https://techcrunch.com/2026/08/26/viral-ai-startup-instinct-has-raised-350-million-at-a-2-5-billion-valuation/) — checked 2026-10-21

---

# IronClaw

> A Rust personal assistant emphasizing credential protection and isolated tool execution.

- **Maker:** NEAR AI
- **Catalogue type:** Self-hosted runtime
- **Website:** https://www.ironclaw.com
- **Source code:** https://github.com/nearai/ironclaw
- **Also known as:** unknown — no other names recorded
- **Evidence:** 7 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/ironclaw/
- **This record:** https://agentlist.wiki/agents/ironclaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1, 5 |
| Included in another plan | ChatGPT subscription through Codex provider | documented | 5 |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT OR Apache-2.0 | documented | 1 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | Startup | documented | 7 |
| Launch date | Feb 23, 2026 | documented | 7 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [7]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 23, 2026 · Public launch [7]

**Launch history**

- 2026-02-23 · Public launch [7]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — WASM tool isolation and credential injection that keeps raw secrets outside tool code. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Cron, events, webhooks, and heartbeat. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Web interface and terminal on macOS, Linux, and Windows; native desktop app not specified. [1]
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — REPL, web gateway, webhooks, Telegram, Slack. [1]
- **Messaging services** `documented` — Telegram, Slack [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — Self-hosted Chat Completions and Responses API routes, described in implementation docs. [4]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — NEAR AI, Anthropic, OpenAI, Gemini, Copilot, Ollama, Bedrock, and compatible endpoints. [5]
- **Bring your own model** `documented` — Own keys for Anthropic, OpenAI, Gemini and other providers; Ollama and compatible endpoints. [5]
- **Included in another plan** `documented` — Can use a ChatGPT subscription through the OpenAI Codex provider. [5]
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1, 5]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT OR Apache-2.0 [1]
- **Where it runs** `documented` — Self-hosted; background service and browser interface. [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Built-in encrypted credential store; website-password autofill is not specified. [3]
- **Approval before actions** `documented` — Scoped approval requests, durable denials, and configurable always-allow policies. [6]
- **What it remembers** `documented` — Persistent workspace, identity files, hybrid full-text/vector search. [1]
- **Edit & export your data** `limited` (partial or conditional only) — Native installers and editable local configuration. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — WASM capabilities, credential injection, endpoint allowlists, Docker isolation. [1]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT OR Apache-2.0 open-source license; Own keys for Anthropic, OpenAI, Gemini and other providers; Ollama and compatible endpoints. | Usage limits and running costs depend on your chosen provider and deployment. | 1, 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web interface and terminal on macOS, Linux, and Windows; native desktop app not specified. | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Telegram | Available | REPL, web gateway, webhooks, Telegram, Slack. | 1 |
| Slack | Available | REPL, web gateway, webhooks, Telegram, Slack. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-29 · Release · [IronClaw 1.4.1](https://github.com/nearai/ironclaw/releases/tag/ironclaw-v1.4.1) · NEAR AI — Stable promotion of a release candidate, fixing Google OAuth activation so Gmail and Calendar extensions work when the operator supplies their own OAuth client, and taking a Wasmtime security update.

## Sources

1. [IronClaw README](https://raw.githubusercontent.com/nearai/ironclaw/main/README.md) — checked 2026-10-21
2. [IronClaw official repository](https://github.com/nearai/ironclaw) — checked 2026-10-21
3. [IronClaw encrypted secret store](https://github.com/nearai/ironclaw/blob/main/crates/substrates/ironclaw_secrets/README.md) — checked 2026-10-21
4. [IronClaw OpenAI-compatible API contract](https://github.com/nearai/ironclaw/blob/main/docs/internal/reborn/contracts/openai-compatible-api.md) — checked 2026-10-21
5. [IronClaw inference providers](https://github.com/nearai/ironclaw/blob/main/docs/capabilities/llm-providers.md) — checked 2026-10-21
6. [IronClaw approval system](https://github.com/nearai/ironclaw/blob/main/crates/kernel/ironclaw_approvals/README.md) — checked 2026-10-21
7. [NEAR AI introduces IronClaw at NEARCON](https://near.ai/blog/near-ai-launches-ironclaw-confidential-gpu-marketplace-and-multimodal-confidential-inference) — checked 2026-10-21

---

# Khoj

> A self-hosted personal AI that indexes your own documents and the web, answers questions over them, and runs scheduled research jobs.

- **Maker:** Khoj AI
- **Catalogue type:** Self-hosted runtime
- **Website:** https://khoj.dev
- **Source code:** https://github.com/khoj-ai/khoj
- **Also known as:** unknown — no other names recorded
- **Evidence:** 13 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/khoj/
- **This record:** https://agentlist.wiki/agents/khoj.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 3, 12 |
| Price & usage costs | Free; no hosted tier | documented | 12 |
| Included in another plan | Self-host only | documented | 12 |
| Where it runs | Self-hosted | documented | 3 |
| Open source & license | AGPL-3.0 | documented | 10 |
| Controls your computer | No | documented | 5 |
| Its own computer | Own hardware · Watch only | documented | 5 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 10, 12 |
| Remote shut-off | Not possible | documented | 3 |
| What you keep | Partial | limited | 3 |
| Company type | Startup | documented | 1, 13 |
| Launch date | Aug 15, 2022 | documented | 11 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [1, 13]
- **Company stage** `documented` — Y Combinator, Summer 2023 [1, 13]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Aug 15, 2022 · Open-source release · First public release [11]

**Launch history**

- 2022-08-15 · Open-source release · First public release [11]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Indexes your own documents alongside the web and answers over both, with custom agents carrying their own knowledge and personas [1, 2]
- **Computer use** `limited` (partial or conditional only) — Khoj Operator is experimental and drives its own isolated Docker container running an Ubuntu desktop, not the user's machine [5]
- **Controls your computer** `documented` — Operator drives its own Docker container rather than the user's desktop. [5]
- **Its own computer** `documented` — An isolated Ubuntu desktop inside a Docker container on the host you run, observable over VNC. [5]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Operator drives a browser inside its own container; currently supports Anthropic models only [5]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Automations run cron-style scheduled queries and email the results; self-hosting requires configuring an email provider [4]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — A desktop companion app syncs files, and the web app installs as a progressive web app; there are also Obsidian and Emacs plugins [7]
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — Web app, desktop companion, Obsidian and Emacs plugins, plus WhatsApp [6, 7]
- **Messaging services** `documented` — WhatsApp, Email [4, 6]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — API keys are issued per client and used by the desktop, Obsidian and Emacs clients; no public API reference page was found [7]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Any online or local model, including OpenAI, Anthropic and Gemini, plus Ollama, LM Studio and offline GGUF models through llama.cpp [3, 9]
- **Bring your own model** `documented` — Bring your own API key, or run entirely offline models [3, 9]
- **Included in another plan** `documented` — Open-source software with no hosted tier; the vendor cloud was shut down in April 2026 [12]
- **Price & usage costs** `documented` — Free software. The hosted Khoj Cloud closed on 15 April 2026 and no paid tier is published [12]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — AGPL-3.0 software you host yourself, so it keeps running if the company stops; the vendor's own hosted service has already closed while the project continues. [10, 12]
- **Remote shut-off** `documented` — Self-hosted with no activation or licence check described. [3]
- **What you keep** `limited` (partial or conditional only) — State lives in a Postgres database and a local directory on your own host, but no documented export or restore path exists. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — AGPL-3.0 [10]
- **Where it runs** `documented` — Self-hosted by pip or Docker on your own machine or server [3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `limited` (partial or conditional only) — Indexed documents and chat history in Postgres with embeddings, plus agents carrying their own knowledge; no user-facing memory browser is documented [3, 8]
- **Edit & export your data** `limited` (partial or conditional only) — Self-hosted state lives in Postgres and a local directory you control; no self-serve export is described [3]
- **Privacy & isolation** `documented` — Self-hosted deployment keeps documents and chat history on your own infrastructure [3, 8]
- **Availability & access** `documented` — Open source and self-hostable; the hosted service is closed [3, 12]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | US$0 | Free software | Full features when self-hosted; Bring your own model keys, or run offline models | You provide hosting and model costs; The hosted Khoj Cloud closed on 15 April 2026 | 3, 12 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Self-hosted web app, installable as a progressive web app. | 3, 7 |
| macOS | Available | Desktop companion app for file sync. | 7 |
| Windows | Available | Desktop companion app for file sync. | 7 |
| Linux | Available | Desktop companion app for file sync. | 7 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | WhatsApp is documented as a Khoj client. | 6 |
| Email | Available | Automations deliver scheduled results by email; delivery is outbound only. | 4 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-03-26 · Release · [Khoj 2.0.0-beta.28](https://github.com/khoj-ai/khoj/releases/tag/2.0.0-beta.28) · Khoj AI — Fixes use of the web app after disabling memories for users who never set a chat model, and stops a deprecation banner showing on the hosted service.

## Sources

1. [Khoj official website](https://khoj.dev) — checked 2026-10-22
2. [Khoj documentation](https://docs.khoj.dev) — checked 2026-10-22
3. [Khoj self-hosting setup](https://docs.khoj.dev/get-started/setup) — checked 2026-10-22
4. [Khoj automations](https://docs.khoj.dev/features/automations) — checked 2026-10-22
5. [Khoj Operator](https://github.com/khoj-ai/khoj/blob/master/src/khoj/processor/operator/README.md) — checked 2026-10-22
6. [Khoj WhatsApp client](https://docs.khoj.dev/clients/whatsapp) — checked 2026-10-22
7. [Khoj desktop client](https://docs.khoj.dev/clients/desktop) — checked 2026-10-22
8. [Khoj privacy](https://docs.khoj.dev/privacy) — checked 2026-10-22
9. [Khoj with Ollama](https://docs.khoj.dev/advanced/ollama) — checked 2026-10-22
10. [Khoj licence](https://github.com/khoj-ai/khoj/blob/master/LICENSE) — checked 2026-10-22
11. [Khoj first release](https://github.com/khoj-ai/khoj/releases/tag/v0.1.5-alpha) — checked 2026-10-22
12. [Khoj Cloud shutdown notice](https://app.khoj.dev/) — checked 2026-10-22
13. [Khoj at 10,000 stars, noting Y Combinator](https://blog.khoj.dev/posts/10k-stars-in/) — checked 2026-10-22

---

# Kimi Claw

> Managed OpenClaw deployment and a Kimi chat connection for existing OpenClaw instances.

- **Maker:** Kimi
- **Catalogue type:** Hosted agent
- **Website:** https://www.kimi.com/bot
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 5 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/kimi-claw/
- **This record:** https://agentlist.wiki/agents/kimi-claw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · paid plan | documented | 1 |
| Price & usage costs | CNY199/mo (China) | documented | 3 |
| Included in another plan | China Pro / Max / Ultra; eligible legacy plans | documented | 1, 4 |
| Where it runs | Kimi cloud or self | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Partly | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 5 |
| Company type | unknown | unknown | — |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — One-click hosted OpenClaw, or connect an existing self-hosted OpenClaw to Kimi. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Kimi Web Search is included; interactive browser control is not specified. [1]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Scheduled tasks through /cron. [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — Kimi, Feishu, WeCom, WeChat, and others. [1]
- **Messaging services** `documented` — WeChat, WeCom, Feishu [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Kimi K2.6 default; K3 configuration documented. [1]
- **Bring your own model** `limited` (partial or conditional only) — Kimi model access is included; custom model-provider/API-key support is not specified. [1]
- **Included in another plan** `documented` — Dedicated Kimi Claw is included with China's current Pro, Max and Ultra plans. Eligible legacy Allegretto and higher subscriptions continue; features draw on shared membership credits. [1, 4]
- **Price & usage costs** `documented` — China plans with dedicated Kimi Claw: Pro CNY199/month, Max CNY699/month, Ultra CNY1,399/month; annual discounts available. The deployed cloud host also consumes about 0.6% of membership credits daily. [3]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — The hosted deployment depends on Kimi's cloud, but an existing self-hosted OpenClaw instance can connect and would survive it. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A request for a copy in a structured, machine-readable format, or transmission to another controller. No retention period after deletion is stated. [5]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Kimi cloud; existing self-hosted instances can connect. [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Long-term OpenClaw memory. [1]
- **Edit & export your data** `documented` — Links existing OpenClaw; hosted-state export unverified. [1]
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `documented` — Available with Allegretto or higher membership. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Pro · China | CNY199/month | Monthly, or CNY1,908/year | Dedicated Kimi Claw and Goal Mode; 4× Agent credits; Kimi Code and Agent Swarm | Monthly shared credits do not roll over; Cloud host consumes about 0.6% of membership credits/day | 3, 4 |
| Max · China | CNY699/month | Monthly, or CNY6,708/year | Dedicated Kimi Claw; 14× Agent credits; Four concurrent Agent tasks | Monthly shared credit pool; Daily host credit charge also applies | 3, 4 |
| Ultra · China | CNY1,399/month | Monthly, or CNY13,428/year | Dedicated Kimi Claw; 28× Agent credits; Four concurrent Agent tasks; 50GB project storage | Monthly shared credit pool; Daily host credit charge also applies | 3, 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

unknown — no native app or web client is recorded.

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WeChat | Available | Kimi, Feishu, WeCom, WeChat, and others. | 1 |
| WeCom | Available | Kimi, Feishu, WeCom, WeChat, and others. | 1 |
| Feishu | Available | Kimi, Feishu, WeCom, WeChat, and others. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Ecosystem

- Built on: OpenClaw → Kimi Claw — Kimi documents deploying OpenClaw in its cloud and connecting existing OpenClaw installations. (https://www.kimi.com/en/help/kimi-claw/overview)

## Sources

1. [Kimi Claw cloud deployment and legacy eligibility](https://www.kimi.com/en/help/kimi-claw/overview) — checked 2026-10-21
2. [Kimi Claw terminal commands](https://www.kimi.com/en/help/kimi-claw/concepts) — checked 2026-10-21
3. [Kimi China membership prices and cloud-host charges](https://www.kimi.com/en/help/membership/membership-pricing) — checked 2026-10-21
4. [Kimi China membership benefits and legacy plans](https://www.kimi.com/en/help/membership/membership-overview) — checked 2026-10-21
5. [Kimi Claw privacy agreement](https://www.kimi.com/user/agreement/en/kimiClawPrivacy) — checked 2026-10-22

---

# Letta

> An open-source stateful agent harness whose memory is a git-backed filesystem you can read, edit and version, runnable on your own machine or on Letta Cloud.

- **Maker:** Letta
- **Catalogue type:** Self-hosted runtime
- **Website:** https://www.letta.com
- **Source code:** https://github.com/letta-ai/letta-code
- **Also known as:** MemGPT, Letta Code
- **Evidence:** 16 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/letta/
- **This record:** https://agentlist.wiki/agents/letta.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 3, 10 |
| Price & usage costs | Free; Cloud from US$20/mo | documented | 10 |
| Included in another plan | Open-source harness, optionally with a Letta Cloud subscription | documented | 10 |
| Where it runs | Self-hosted or Letta Cloud | documented | 3 |
| Open source & license | Apache-2.0 | documented | 14 |
| Controls your computer | Limited | documented | 9 |
| Its own computer | Cloud | documented | 9 |
| Works while your computer is off | Partly | limited | 6 |
| If the vendor stops | Keeps working | documented | 3, 14 |
| Remote shut-off | Not possible | documented | 3 |
| What you keep | Self-serve | documented | 4 |
| Company type | Research group | documented | 1 |
| Launch date | Oct 26, 2023 | documented | 14 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Research group [1]
- **Company stage** `documented` — AI research lab in San Francisco, founded by the creators of MemGPT at UC Berkeley [1]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Oct 26, 2023 · Open-source release · First release as MemGPT [14]

**Launch history**

- 2023-10-26 · Open-source release · First release as MemGPT [14]
- 2024-09-23 · Major update · Renamed from MemGPT to Letta [15]
- 2025-12-16 · Major update · Letta Code, the memory-first agent harness [16]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Memory is MemFS, a git-backed filesystem of Markdown files the agent and the user can both inspect and edit, with every change committed [4, 5]
- **Computer use** `limited` (partial or conditional only) — A computer in Letta is an execution environment with a shell and filesystem, on the current machine, a cloud sandbox, or a machine you bring; it is not graphical desktop control [9]
- **Controls your computer** `documented` — Runs a shell and filesystem on the current machine, which is execution rather than graphical desktop control. [9]
- **Its own computer** `documented` — Cloud sandboxes, or a machine you bring, with agents able to move between them. [9]
- **Works while your computer is off** `limited` (partial or conditional only) — Cloud schedules run on a server-side timer; local schedules run only while a Letta session is open on that computer. [6]
- **Browser control** `limited` (partial or conditional only) — A built-in browser-use skill appears in the changelog, but there is no dedicated documentation page for it; web search and page fetching are documented for cloud agents [13]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Schedules run one-off or recurring prompts on cron, server-side on Cloud and only while a session is open locally; sleep-time subagents review recent conversations and consolidate memory in the background [5, 6]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — No native desktop app; Letta Desktop is the web app installed as a progressive web app, and the former Electron app was retired in October 2026 [12]
- **iPhone & Android apps** `documented` — No official iOS or Android app; the documentation recommends the mobile web app [13]
- **Web & messaging** `documented` — Web app, desktop PWA and a command-line interface, plus messaging channels on the local backend [8, 12]
- **Messaging services** `documented` — WhatsApp, Telegram, Signal, Slack, Discord [8]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Public REST API at api.letta.com for Cloud agents, a TypeScript agent SDK, generated client SDKs, and a hosted MCP server [11, 13]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Model-agnostic across roughly 35 providers including Anthropic, OpenAI, Gemini, Bedrock, Mistral, DeepSeek and Groq, with local inference through Ollama, LM Studio and llama.cpp; the model can be changed mid-conversation [7]
- **Bring your own model** `documented` — Bring your own API keys on every plan, including the free tier [7, 10]
- **Included in another plan** `documented` — Open-source harness, optionally with a Letta Cloud subscription [10]
- **Price & usage costs** `documented` — Free software. Letta Cloud is free for three agents, Pro US$20/month, an API plan at US$20/month plus usage, Teams Pro US$20/seat/month and custom Enterprise pricing [10]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Apache-2.0 software that runs entirely on your own machine with no account required, so it keeps running if the company stops. [3, 14]
- **Remote shut-off** `documented` — Run locally, there is no account requirement or activation check described. [3]
- **What you keep** `documented` — Memory is a git repository on your own disk with full history; the documentation notes you are responsible for backing it up. [4]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — Apache-2.0 [14]
- **Where it runs** `documented` — Self-hosted locally or by Docker, or hosted on Letta Cloud; run locally, all agent state stays on the device and no account is required [3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — A context repository of Markdown files holding persona, facts about the user, project facts, workflow rules and agent skills; files at the root load every turn and subdirectories load on demand [4, 5]
- **Edit & export your data** `documented` — Memory is a real git repository with full version history; local agents commit on the user's own machine and cloud agents can sync to GitHub [4]
- **Privacy & isolation** `documented` — Run locally, agent state including messages, memory and provider connections stays on the device; Enterprise adds role-based access control and SSO [3, 10]
- **Availability & access** `documented` — Open source and installable from npm; Letta Cloud has open sign-up with a free tier [3, 10]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free tier on Letta Cloud | Three stateful agents; 5 MiB of storage per memory repository; Bring your own API keys | Limited use of Letta Auto | 10 |
| Pro | US$20/month | Monthly | Up to 20 stateful agents; 50 MiB per memory repository; Weekly and monthly Letta Auto quota | Pay-as-you-go overage | 10 |
| API | US$20/month | Monthly plus usage | Unlimited agents; API key authentication | US$0.10 per active agent per month; US$0.00015 per second of tool execution | 10 |
| Teams Pro | US$20/user/month | Per seat, monthly | Shared agents with access control; Weekly and monthly Letta Auto quota | Priced per seat | 10 |
| Enterprise | Custom | Priced on request | Role-based access control; SAML and OIDC single sign-on; Increased quotas and dedicated support | Volume-based pricing on request | 10 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web app at chat.letta.com, installable as a progressive web app. | 12 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | Self-hosted messaging channel on the local backend. | 8 |
| Telegram | Available | Self-hosted messaging channel on the local backend. | 8 |
| Signal | Available | Self-hosted messaging channel on the local backend. | 8 |
| Slack | Available | A self-hosted channel, and a first-party Slack integration for Cloud agents. | 8 |
| Discord | Available | Self-hosted messaging channel on the local backend. | 8 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-22 · Release · [Letta v0.34.21](https://github.com/letta-ai/letta-code/releases/tag/v0.34.21) · Letta — Carries MCP originator tokens through to tool calls for user-scoped connections, and sends the selected reasoning effort for xAI model handles.

## Sources

1. [Letta official website](https://www.letta.com/) — checked 2026-10-22
2. [Letta documentation](https://docs.letta.com/) — checked 2026-10-22
3. [Letta self-hosting](https://docs.letta.com/self-hosting) — checked 2026-10-22
4. [Letta MemFS memory filesystem](https://docs.letta.com/concepts/memfs) — checked 2026-10-22
5. [Letta memory configuration](https://docs.letta.com/configuration/memory) — checked 2026-10-22
6. [Letta schedules](https://docs.letta.com/configuration/schedules) — checked 2026-10-22
7. [Letta model providers](https://docs.letta.com/configuration/models) — checked 2026-10-22
8. [Letta messaging channels](https://docs.letta.com/self-hosting/channels) — checked 2026-10-22
9. [Letta computers and execution environments](https://docs.letta.com/platform/computers) — checked 2026-10-22
10. [Letta pricing](https://docs.letta.com/pricing) — checked 2026-10-22
11. [Letta API reference](https://docs.letta.com/api/resources/agents) — checked 2026-10-22
12. [Letta Desktop](https://docs.letta.com/platform/letta-desktop) — checked 2026-10-22
13. [Letta frequently asked questions](https://docs.letta.com/reference/faq) — checked 2026-10-22
14. [Letta Code repository](https://github.com/letta-ai/letta-code) — checked 2026-10-22
15. [Announcing Letta](https://www.letta.com/blog/announcing-letta) — checked 2026-10-22
16. [Letta Code](https://www.letta.com/blog/letta-code) — checked 2026-10-22

---

# Lindy

> A personal and team assistant with editable memory files, reusable skills, and recurring routines.

- **Maker:** Lindy
- **Catalogue type:** Hosted agent
- **Website:** https://www.lindy.ai
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 10 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/lindy/
- **This record:** https://agentlist.wiki/agents/lindy.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | $49.99/user/mo | documented | 4 |
| Included in another plan | Standalone Lindy subscription with pooled credits | documented | 4 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 10 |
| Company type | Startup | documented | 9 |
| Launch date | Mar 21, 2023 | documented | 9 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [9]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Mar 21, 2023 · Announcement [9]

**Launch history**

- 2023-03-21 · Announcement [9]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Teach tasks through a recording with Watch and Learn; reusable skills and memory shared with your team [7]
- **Computer use** `limited` (partial or conditional only) — Browser automation and shell/code tools; full local-desktop control is not specified [1, 5]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Can drive websites when integrations are unavailable; paid-plan access [1, 4]
- **Voice & ambient listening** `documented` — Voice memos, meeting capture and phone calls; calls currently use US numbers [7, 8]
- **Background work & triggers** `documented` — Recurring routines on a schedule [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Web app and Chrome extension; native desktop app not specified [7]
- **iPhone & Android apps** `limited` (partial or conditional only) — iMessage/SMS and Slack access; native mobile app not specified in current product guide [7]
- **Web & messaging** `documented` — Web and connected work tools [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — Connect hosted MCP tools; public agent-control API not specified [6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Multiple major models, selectable for each task [4]
- **Bring your own model** `limited` (partial or conditional only) — Choose a supplied model per task; use of your own inference keys is not specified [4]
- **Included in another plan** `documented` — Standalone Lindy subscription with pooled credits [4]
- **Price & usage costs** `documented` — Seven-day trial with US$50 credits. Team starts at US$49.99 per active user/month for 5,000 pooled credits; top-ups cost US$10 per 1,000 credits. [4]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Review, update or deletion by emailing a privacy address; no self-serve export, no stated format, and no retention period after cancellation. [10]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Hosted service [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Built-in write-only API-key credentials, private or workspace-shared; general password vault not specified [5]
- **Approval before actions** `documented` — External-impact actions such as email, ticket changes and publishing wait for approval [4]
- **What it remembers** `documented` — User-editable memory in plain files [1]
- **Edit & export your data** `limited` (partial or conditional only) — Memory is stored in editable plain files with version history; full runtime export not specified [1, 7]
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `documented` — Self-serve access [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free trial | US$0 | Seven days | US$50 in trial credits; No card required | Credits expire after seven days | 4 |
| Team | From US$49.99/user/month | Monthly; every active user is billed | 5,000 credits per seat, pooled across the team; Model selection, computer use and onboarding | Larger credit tiers available; Monthly credits expire; purchased top-ups do not; Top-ups: US$10/1,000 credits; Work pauses when the balance is depleted | 4 |
| Enterprise | Custom | Custom contract | Team features plus SSO, SCIM and audit logs; HIPAA support with BAA; dedicated support | Custom credit allowance | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web app and Chrome extension; native desktop app not specified | 7 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.lindy.ai/changelog — Dated changelog with per-entry anchors (checked 2026-10-22)

- 2026-10-18 · Release · [New integrations and quick questions](https://www.lindy.ai/changelog#2026-10-18-integrations) · Lindy — Adds connectors including Airtable, Intercom, Google Analytics and ClickUp, plus decision forms that ask the user a quick question mid-task.
- 2026-10-16 · Release · [Booking links](https://www.lindy.ai/changelog#2026-10-16-booking-links) · Lindy — Shareable scheduling links with configurable duration, hours, breaks and video app, syncing to Google and Outlook calendars.

## Sources

1. [Lindy · Official product page](https://www.lindy.ai/) — checked 2026-10-21
2. [Memory · Lindy](https://www.lindy.ai/product/memory) — checked 2026-10-21
3. [Routines · Lindy](https://www.lindy.ai/product/routines) — checked 2026-10-21
4. [Lindy current plans, seat billing and credits](https://www.lindy.ai/pricing) — checked 2026-10-21
5. [Lindy credentials: API-key storage and sharing](https://docs.lindy.ai/integrations/credentials) — checked 2026-10-21
6. [Connect MCP servers to Lindy](https://docs.lindy.ai/integrations/mcp) — checked 2026-10-21
7. [Lindy current product capabilities](https://www.lindy.ai/product) — checked 2026-10-21
8. [Lindy phone calls](https://www.lindy.ai/product/phone) — checked 2026-10-21
9. [Lindy’s first product announcement](https://www.lindy.ai/blog/meet-lindy) — checked 2026-10-21
10. [Lindy privacy policy](https://www.lindy.ai/privacy) — checked 2026-10-22

---

# Lucas

> A proactive assistant in iMessage and WhatsApp that follows up on life admin and completes browser errands, with an encrypted credential vault.

- **Maker:** Lucas AI
- **Catalogue type:** Hosted agent
- **Website:** https://meetlucas.ai/
- **Source code:** unknown — no public repository recorded
- **Also known as:** meetlucasai, Meet Lucas
- **Evidence:** 7 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/lucas/
- **This record:** https://agentlist.wiki/agents/lucas.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Free; $20 or $100/mo | documented | 2 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Provider cloud | documented | 3 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | limited | 3 |
| Company type | Startup | documented | 7 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [7]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Completes reservations, refunds and other life admin from texts; encrypted Vault keeps logins, cards and ID documents ready for errands. [1, 4]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Runs up to five browser errands concurrently, with a live view and handover when needed. [5]
- **Voice & ambient listening** `documented` — Transcribes WhatsApp voice notes and acts on them. [6]
- **Background work & triggers** `documented` — Proactive reminders, briefings, automation updates and meeting alerts. [6]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `documented` — Uses existing messaging apps; no separate Lucas mobile app is required. [7]
- **Web & messaging** `documented` — iMessage, WhatsApp and a web dashboard. [1, 3]
- **Messaging services** `documented` — iMessage, WhatsApp [1, 3]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Uses third-party AI services including OpenAI, Anthropic and Google. [3]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Free tier; Pro $20/month or $192/year; Max $100/month or $960/year. New accounts get a seven-day paid-feature trial before reverting to Free. [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `limited` (partial or conditional only) — The policy offers access, update, export and deletion of certain account information in-product "where those tools are available", which leaves what can actually be exported unsettled. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed assistant using cloud infrastructure and third-party AI services. [3]
- **Passwords & credentials** `documented` — Built-in Vault encrypts passwords, payment cards and IDs. Browser filling keeps secret fields out of the errand model’s context, with per-entry activity logs. [4]
- **Approval before actions** `documented` — Bookings and payments require approval of the final checkout screenshot; changed prices or details require fresh approval. [5]
- **What it remembers** `documented` — Keeps personal facts, preferences, relationships and routines; ask Lucas to remember, change or forget details. [2, 3]
- **Edit & export your data** `documented` — Manage saved memory and connected accounts; account deletion starts removal subject to backup and legal retention. [3]
- **Privacy & isolation** `documented` — Vault entries use AES-256 with per-user keys; secret fields are encrypted separately and restricted to approved sites. [4]
- **Availability & access** `documented` — Sign up with a mobile number and use iMessage or WhatsApp. [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | $0 | Ongoing free tier | Seven-day paid-feature trial for new accounts | Usage limits apply from day one; exact allowances are shown in the account. | 2 |
| Pro | $20/month or $192/year | Monthly or annual subscription | — | Exact feature and usage allowances are shown in the account. | 2 |
| Max | $100/month or $960/year | Monthly or annual subscription | — | Exact feature and usage allowances are shown in the account. | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Limited availability | Web dashboard for integrations, account settings and Vault; conversations run through messaging. | 3, 4 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | iMessage, WhatsApp and a web dashboard. | 1, 3 |
| WhatsApp | Available | iMessage, WhatsApp and a web dashboard. | 1, 3 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-15 · Release · [Lucas launches an encrypted credential Vault](https://meetlucas.ai/updates/2026-09-15-introducing-vault/) — Vault stores logins, cards and IDs for browser errands while keeping secret fields out of the agent model’s context; purchases still require approval.

## Sources

1. [Lucas product overview](https://meetlucas.ai/) — checked 2026-10-21
2. [Lucas FAQ and pricing](https://meetlucas.ai/faq/) — checked 2026-10-21
3. [Lucas privacy policy](https://meetlucas.ai/privacy-policy/) — checked 2026-10-21
4. [Introducing Vault](https://meetlucas.ai/updates/2026-09-15-introducing-vault/) — checked 2026-10-21
5. [Parallel errands and flight booking](https://meetlucas.ai/updates/2026-09-14-parallel-errands-and-flight-booking/) — checked 2026-10-21
6. [Lucas product updates](https://meetlucas.ai/updates/) — checked 2026-10-21
7. [About Lucas](https://meetlucas.ai/about/) — checked 2026-10-21

---

# Manus

> A general agent with its own cloud computer and projects that carry instructions and files between tasks, plus scheduled and event-triggered automations.

- **Maker:** Manus AI
- **Catalogue type:** Hosted agent
- **Website:** https://manus.im
- **Source code:** unknown — no public repository recorded
- **Also known as:** Manus AI
- **Evidence:** 12 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/manus/
- **This record:** https://agentlist.wiki/agents/manus.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1, 3 |
| Price & usage costs | Free; from US$20/mo | documented | 8, 9 |
| Included in another plan | Standalone Manus subscription | documented | 9 |
| Where it runs | Provider cloud + your computer | documented | 4, 5 |
| Open source & license | unknown | unknown | — |
| Controls your computer | Yes · Mac, Windows | documented | 5 |
| Its own computer | Cloud | documented | 4 |
| Works while your computer is off | Keeps working | documented | 4 |
| If the vendor stops | Stops | limited | 4 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 11 |
| Company type | Startup | documented | 2, 12 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [2, 12]
- **Company stage** `documented` — Independent; headquartered in Singapore [2, 12]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

- 2026-09-28 · Major update · Manus 2.0 on web, desktop and mobile [3]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — A persistent cloud computer that keeps running between tasks, plus projects holding a master instruction and a knowledge base [4, 6]
- **Computer use** `documented` — Runs a dedicated Ubuntu cloud machine, and the desktop app's My Computer can read authorised local folders, run shell commands and launch applications with explicit approval [4, 5]
- **Controls your computer** `documented` — My Computer in the desktop app reads authorised folders, runs shell commands and launches applications, each with explicit approval. [5]
- **Its own computer** `documented` — A dedicated Ubuntu cloud computer runs bots and scripts around the clock, reachable over SSH or a web terminal. [4]
- **Works while your computer is off** `documented` — The cloud computer runs around the clock independently of the user's device. [4]
- **Browser control** `documented` — Browses and acts on the web from its own machine as part of task execution [3, 4]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Automations run on a schedule down to five-minute intervals or on cron, and trigger on events from Gmail, Outlook, Notion, Google Calendar, GitHub, Shopify, RSS and webhooks [7]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native macOS and Windows apps [3, 5]
- **iPhone & Android apps** `documented` — Native iOS and Android apps [3]
- **Web & messaging** `documented` — Web, desktop and mobile apps, plus Slack and email [3, 7]
- **Messaging services** `documented` — Slack, Email [7]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Public REST API with endpoints for tasks, projects, files, webhooks, skills and agents [10]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `limited` (partial or conditional only) — Fully managed; personal model keys are not described [8]
- **Included in another plan** `documented` — Standalone Manus subscription [9]
- **Price & usage costs** `documented` — Free tier with 300 daily credits; Pro from US$20/month; a higher Pro tier at US$40/month; Team from US$20/seat/month. Annual billing is discounted 17% [8, 9]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [4]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A data backup tool produces task and account backups, downloadable or exported to Google Drive or OneDrive. [11]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Provider-hosted cloud sandbox and a dedicated always-on cloud computer; the desktop app also executes locally [4, 5]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Local file and application access through My Computer requires explicit approval [5]
- **What it remembers** `documented` — Projects hold a master instruction and a knowledge base of files; the cloud computer keeps a persistent disk across tasks [4, 6]
- **Edit & export your data** `documented` — A backup tool exports task and account data for download or to Google Drive or OneDrive [11]
- **Privacy & isolation** `limited` (partial or conditional only) — Team plans add single sign-on and data privacy controls [9]
- **Availability & access** `documented` — Open sign-up on the web, desktop and mobile [1, 3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free tier | 300 credits daily; One concurrent task | Plan credits reset each cycle | 8, 9 |
| Pro | From US$20/month | Monthly; 17% discount billed annually | About 4,000 credits a month; Advanced research, website deployment and slides | A higher Pro tier at US$40/month adds about 8,000 credits | 9 |
| Team | From US$20/user/month | Monthly; 17% discount billed annually | Single sign-on and data privacy controls; Team analytics and shared templates | Priced per seat | 9 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Manus 2.0 is available on the web. | 3 |
| macOS | Available | Native macOS desktop app with local file and application access. | 3, 5 |
| Windows | Available | Native Windows desktop app with local file and application access. | 3, 5 |
| iOS | Available | Native iOS app. | 3 |
| Android | Available | Native Android app. | 3 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Slack | Available | Mention Manus in Slack to post updates and assign tasks. | 7 |
| Email | Available | Mail Manus to turn an email into an action; Gmail and Outlook also act as automation triggers. | 7 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://manus.im/blog — Product blog with dated posts (checked 2026-10-22)

- 2026-10-01 · Release · [Video Editor in Manus 2.0](https://manus.im/blog/introducing-video-editor) · Manus — A desktop timeline editor that keeps generated video as separate layers for captions, music and effects, alongside prompt-to-video generation.
- 2026-10-01 · Release · [Game Dev mode](https://manus.im/blog/introducing-game-developer) · Manus — A game-building mode with a live panel for tweaking parameters, swapping assets and hosting multiplayer.
- 2026-09-29 · Release · [Manus Flex brings your own API key](https://manus.im/blog/introducing-manus-flex) · Manus — Lets people use their own inference provider keys in the Manus workspace, billed by that provider rather than in Manus credits.

## Sources

1. [Manus official website](https://manus.im/) — checked 2026-10-22
2. [About Manus](https://manus.im/about) — checked 2026-10-22
3. [Introducing Manus 2.0](https://manus.im/blog/introducing-manus-2-0) — checked 2026-10-22
4. [Manus Cloud Computer](https://manus.im/blog/manus-cloud-computer) — checked 2026-10-22
5. [Manus desktop app and My Computer](https://manus.im/docs/features/desktop) — checked 2026-10-22
6. [Manus projects](https://manus.im/docs/features/projects) — checked 2026-10-22
7. [Manus automations](https://manus.im/docs/automations) — checked 2026-10-22
8. [Manus plans and credits](https://manus.im/docs/introduction/plans) — checked 2026-10-22
9. [Manus membership pricing](https://help.manus.im/en/articles/11711111-what-is-the-current-membership-pricing-for-manus) — checked 2026-10-22
10. [Manus API documentation](https://open.manus.ai/docs) — checked 2026-10-22
11. [Backing up your Manus data](https://help.manus.im/en/articles/16147892-service-change-overview-how-to-back-up-your-data) — checked 2026-10-22
12. [Manus resumes independent operations](https://manus.im/blog/manus-resumes-independent-operations) — checked 2026-10-22

---

# Martin

> A personal assistant reached by text, phone call or email that manages calendar and inbox, keeps notes and to-dos, and takes proactive actions.

- **Maker:** DCWU, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://www.trymartin.com
- **Source code:** unknown — no public repository recorded
- **Also known as:** Try Martin
- **Evidence:** 9 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/martin/
- **This record:** https://agentlist.wiki/agents/martin.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | From US$21/mo billed yearly | documented | 2 |
| Included in another plan | Standalone Martin subscription | documented | 2 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | No | limited | 3 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | unknown | unknown | — |
| Company type | Startup | documented | 8 |
| Launch date | Aug 17, 2023 | documented | 8 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [8]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Aug 17, 2023 · Public launch · iOS app first released [8]

**Launch history**

- 2023-08-17 · Public launch · iOS app first released [8]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Reached by text, phone call or email rather than a new interface, with wake-up calls and copy-on-an-email scheduling [1, 4]
- **Computer use** `limited` (partial or conditional only) — Acts through connected accounts; control of a computer is not described [3]
- **Controls your computer** `limited` (partial or conditional only) — Nothing is described as running on the user's computer; Martin acts through connected accounts. [3]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Phone calls in both directions, including spoken wake-up calls [1, 4]
- **Background work & triggers** `documented` — Background tasks including wake-up calls and reminders; paid tiers are sold on how many proactive actions they allow [2, 4, 5]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — No native desktop app is described [1]
- **iPhone & Android apps** `documented` — Native iPhone and iPad app on the App Store [8]
- **Web & messaging** `documented` — Text messages, phone calls, email, the iOS app and a web app [1, 8]
- **Messaging services** `documented` — SMS, Slack, Email [1, 3, 7]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — Plans distinguish basic from most advanced AI models; the providers are not named [2]
- **Bring your own model** `limited` (partial or conditional only) — Fully managed; personal model keys are not described [2]
- **Included in another plan** `documented` — Standalone Martin subscription [2]
- **Price & usage costs** `documented` — Basic US$21/month billed yearly or US$35 monthly; Pro US$30/month billed yearly or US$49 monthly, with seven days free [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Provider-hosted service reached through messaging, calls and apps [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Keeps notes and to-dos as a productivity suite alongside calendar and inbox context [1, 3]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `documented` — Open sign-up with a seven-day free trial [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Basic | US$21/month billed yearly | US$35/month billed monthly; seven days free | Basic AI models; Two tasks at a time | Limited proactive actions | 2 |
| Pro | US$30/month billed yearly | US$49/month billed monthly; seven days free | Most advanced AI models; Unlimited proactive actions; Unlimited custom tasks | Business pricing is listed separately | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web application at app.trymartin.com. | 1 |
| iOS | Available | Martin: Personal Assistant on the App Store, supporting iPhone and iPad. | 8 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| SMS | Available | Text Martin to give it instructions. | 1 |
| Slack | Available | Slack is listed among Martin's integrations. | 3 |
| Email | Available | Email Martin, or copy it on an email to schedule. | 1, 7 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.trymartin.com/blog — Product blog, dated, with nothing published since April 2025 (checked 2026-10-22)

- 2025-04-23 · Release · [Apps, Notes, v2 to-dos and more languages](https://www.trymartin.com/blog/introducing-martin-apps) · Martin — Adds an apps surface, Notes, a rebuilt to-do system and more supported languages. The product blog has published nothing since.

## Sources

1. [Martin official website](https://www.trymartin.com/) — checked 2026-10-22
2. [Martin pricing](https://www.trymartin.com/pricing) — checked 2026-10-22
3. [Martin documentation](https://docs.trymartin.com/) — checked 2026-10-22
4. [Martin wake-up calls](https://docs.trymartin.com/background-tasks/wake-up-calls) — checked 2026-10-22
5. [Martin reminders](https://docs.trymartin.com/capabilities/reminders) — checked 2026-10-22
6. [Martin calendar integration](https://docs.trymartin.com/integrations/calendar) — checked 2026-10-22
7. [Martin inbox integration](https://docs.trymartin.com/integrations/inbox) — checked 2026-10-22
8. [Martin on the App Store](https://apps.apple.com/us/app/martin/id6453473083) — checked 2026-10-22
9. [Martin privacy policy](https://info.trymartin.com/privacy) — checked 2026-10-22

---

# MaxClaw

> A managed cloud assistant built on OpenClaw with integrated tools and expert agents.

- **Maker:** MiniMax
- **Catalogue type:** Hosted agent
- **Website:** https://agent.minimax.io/max-claw
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 3 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/maxclaw/
- **This record:** https://agentlist.wiki/agents/maxclaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | $19/mo | documented | 1 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 3 |
| Company type | Public company | documented | 2 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Public company [2]
- **Company stage** `documented` — Public [2]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — 50 GB cloud workspace plus custom expert agents built with natural language and visual tools. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Built-in browser actions for multi-step tasks in the hosted agent. [1]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Scheduled monitoring, recurring data collection, and automatic reports. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — Web, mobile, Telegram, Discord, Slack. [1]
- **Messaging services** `documented` — Telegram, Slack, Discord [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — MiniMax M2.7, according to product FAQ. [1]
- **Bring your own model** `limited` (partial or conditional only) — MiniMax model access is included without separate API setup; own-model support is not specified. [1]
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — $19/month is listed on the official MaxClaw page. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A written request to the data protection officer, and the policy states a reasonable fee will be charged for a copy of personal data. No retention period after deletion is stated. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — MiniMax-managed cloud, continuously available. [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Long-term memory and cloud workspace. [1]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `documented` — Available as a hosted subscription. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| MaxClaw | US$19/month | Monthly listed rate | Hosted OpenClaw with MiniMax M2.7; 50GB cloud workspace for new users; Browser tools, integrations and automatic maintenance | Numerical inference allowance and overage rates not listed | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web, mobile, Telegram, Discord, Slack. | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Telegram | Available | Web, mobile, Telegram, Discord, Slack. | 1 |
| Slack | Available | Web, mobile, Telegram, Discord, Slack. | 1 |
| Discord | Available | Web, mobile, Telegram, Discord, Slack. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Ecosystem

- Built on: OpenClaw → MaxClaw — MiniMax hosts OpenClaw as MaxClaw, with managed setup, model access, and a cloud workspace. (https://agent.minimax.io/activity/max-claw)

## Sources

1. [MaxClaw official hosted plan](https://agent.minimax.io/activity/max-claw) — checked 2026-10-21
2. [MiniMax investor relations](https://ir.minimax.io/) — checked 2026-10-21
3. [MiniMax Agent privacy policy](https://agent.minimax.io/doc/en/privacy-policy.html) — checked 2026-10-22

---

# Moltis

> A single-binary Rust agent server you run on your own hardware, with durable memory, sandboxed tool execution and more than ten chat channels.

- **Maker:** Fabien Penso
- **Catalogue type:** Self-hosted runtime
- **Website:** https://moltis.org
- **Source code:** https://github.com/moltis-org/moltis
- **Also known as:** unknown — no other names recorded
- **Evidence:** 9 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/moltis/
- **This record:** https://agentlist.wiki/agents/moltis.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 4 |
| Price & usage costs | Free; you pay models | documented | 1, 3 |
| Included in another plan | Open-source software with no hosted tier | documented | 1 |
| Where it runs | Self-hosted | documented | 4, 5, 6 |
| Open source & license | MIT | documented | 8 |
| Controls your computer | Limited | documented | 2, 7 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 4, 8 |
| Remote shut-off | Not possible | documented | 4 |
| What you keep | Self-serve | documented | 2, 4 |
| Company type | Independent project | documented | 1, 8 |
| Launch date | Feb 3, 2026 | documented | 9 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Independent project [1, 8]
- **Company stage** `documented` — Built by an individual developer [1, 8]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 3, 2026 · Open-source release · First public release [9]

**Launch history**

- 2026-02-03 · Open-source release · First public release [9]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — One Rust binary holding the agent, its memory and more than ten chat channels, with a native desktop client that attaches to the running daemon locally or over SSH [1, 2]
- **Computer use** `limited` (partial or conditional only) — Browser automation runs on the host where the daemon runs, which may be your own machine; desktop or GUI control is not claimed [2, 7]
- **Controls your computer** `documented` — Browser automation runs on the host where the daemon runs, which may be your own machine; desktop control is not claimed. [2, 7]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Built-in browser automation, with SSRF protection on web fetching [2, 7]
- **Voice & ambient listening** `documented` — Telephony and voice are listed among its channels [3]
- **Background work & triggers** `documented` — Cron scheduling, with durable local CalDAV and email connectors [2, 3]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — A native Rust desktop client attaches to a running daemon locally, over HTTPS or through an SSH tunnel, with keychain-stored scoped keys and certificate pinning [2]
- **iPhone & Android apps** `documented` — A mobile progressive web app with push notifications; no native mobile app [2]
- **Web & messaging** `documented` — Web interface, native desktop client, mobile progressive web app and an API, plus messaging channels [2, 3]
- **Messaging services** `documented` — WhatsApp, Telegram, Signal, Slack, Discord, Microsoft Teams, Matrix, Email [2, 3]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — GraphQL API with scoped keys and webhooks [3]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Anthropic, OpenAI, Gemini, DeepSeek, Mistral, Groq, xAI, OpenRouter, Ollama and local models, plus GitHub Copilot and OpenAI Codex over OAuth [3, 4]
- **Bring your own model** `documented` — Bring your own keys, held in an encrypted vault [3, 4]
- **Included in another plan** `documented` — Open-source software with no hosted tier [1]
- **Price & usage costs** `documented` — Free software; no hosted service or paid tier is published [1, 3]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — MIT-licensed single binary you run on your own hardware, so it keeps running if the author stops. [4, 8]
- **Remote shut-off** `documented` — Runs on your own hardware with no activation or licence check described. [4]
- **What you keep** `documented` — Sessions are JSONL files in a local data directory and the web interface exports Markdown. [2, 4]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [8]
- **Where it runs** `documented` — Self-hosted from an install script, Homebrew, Docker or cargo, or deployed to a cloud host you manage [4, 5, 6]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Long-term memory in SQLite with full-text search and vector embeddings, plus session persistence with auto-compaction; you add knowledge by dropping Markdown files into a local memory directory [2, 4]
- **Edit & export your data** `documented` — Sessions are stored as JSONL on your own disk and the web interface exports Markdown [2, 4]
- **Privacy & isolation** `documented` — Tool and code execution is sandboxed in Docker, Podman, Apple Container or WebAssembly, and web fetching is protected against server-side request forgery [7]
- **Availability & access** `documented` — Open source; install by script, Homebrew, Docker or cargo [4]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | US$0 | Free software | Full features when self-hosted; Your own provider keys in an encrypted vault | You provide hosting and model costs | 1, 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web interface served by the daemon. | 2 |
| macOS | Available | Native Rust desktop client; installable by Homebrew. | 2, 4 |
| Linux | Available | Native desktop client and install script. | 2, 4 |
| Mobile app | Available | Mobile progressive web app with push notifications. | 2 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | Listed among Moltis chat channels. | 3 |
| Telegram | Available | Listed among Moltis chat channels. | 3 |
| Signal | Available | Listed among Moltis chat channels. | 3 |
| Slack | Available | Listed among Moltis chat channels. | 3 |
| Discord | Available | Listed among Moltis chat channels. | 3 |
| Microsoft Teams | Available | Listed among Moltis chat channels. | 3 |
| Matrix | Available | Listed among Moltis chat channels. | 3 |
| Email | Available | Email through Gmail and Himalaya connectors. | 2 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-19 · Release · [Moltis 20261019.01](https://github.com/moltis-org/moltis/releases/tag/20261019.01) · Moltis — A dated build published to the project's releases; no written release notes accompany this tag.

## Sources

1. [Moltis official website](https://moltis.org) — checked 2026-10-22
2. [Moltis repository](https://github.com/moltis-org/moltis) — checked 2026-10-22
3. [Moltis documentation](https://docs.moltis.org) — checked 2026-10-22
4. [Moltis quickstart](https://docs.moltis.org/quickstart.html) — checked 2026-10-22
5. [Moltis with Docker](https://docs.moltis.org/docker.html) — checked 2026-10-22
6. [Moltis cloud deployment](https://docs.moltis.org/cloud-deploy.html) — checked 2026-10-22
7. [Moltis security](https://docs.moltis.org/security.html) — checked 2026-10-22
8. [Moltis licence](https://github.com/moltis-org/moltis/blob/main/LICENSE.md) — checked 2026-10-22
9. [Moltis first release](https://github.com/moltis-org/moltis/releases/tag/v0.1.1) — checked 2026-10-22

---

# Muse

> A personal agent that uses a dedicated virtual machine and connected apps to advance ongoing goals.

- **Maker:** Meta
- **Catalogue type:** Hosted agent
- **Website:** https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 9 sources, checked 2026-10-21 to 2026-10-26
- **Profile:** https://agentlist.wiki/agent/muse/
- **This record:** https://agentlist.wiki/agents/muse.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · some regions | limited | 1 |
| Price & usage costs | Free; paid upgrades | limited | 1 |
| Included in another plan | Standalone Muse plans | documented | 1 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | Cloud | documented | 2 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 8 |
| Company type | Public company | documented | 3 |
| Launch date | Sep 8, 2026 | documented | 1 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Public company [3]
- **Company stage** `documented` — Public · NASDAQ: META [3]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Sep 8, 2026 · Public launch [1]

**Launch history**

- 2026-09-08 · Public launch [1]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Tracks long-running goals and event-driven follow-ups; memory files can be read and edited directly. [2]
- **Computer use** `documented` — Dedicated cloud virtual machine with browser, terminal, and files; local desktop control is not specified. [2]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `documented` — A dedicated secure virtual machine with its own browser. Whether you can watch it is not specified. [2]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Cloud browser searches, fills forms, books, and purchases. [2]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Scheduled and event-triggered background work. [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Native Mac app, released 17 September 2026 after the original launch announcement; Windows and Linux are not specified. [7]
- **iPhone & Android apps** `documented` — Native iOS and Android apps, plus WhatsApp access. [1]
- **Web & messaging** `documented` — Muse app and WhatsApp [1]
- **Messaging services** `documented` — WhatsApp [1]
- **Native hardware** `limited` (partial or conditional only) — Meta says Muse is coming to its AI glasses in the coming months, where you say the agent's name and it can act on what you are looking at. Meta's own pages still describe this as coming soon, and the AI glasses release notes do not list it as shipped. Muse's cloud virtual machine is separate from this. [1, 9]
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Muse Spark at launch [1]
- **Bring your own model** `documented` — Muse Spark supplied by the service at launch; BYO models are not specified. [1]
- **Included in another plan** `documented` — Standalone Muse service with free usage and subscription upgrades; no wider paid bundle is listed. [1]
- **Price & usage costs** `limited` (partial or conditional only) — Free for everyday use, with paid subscriptions for higher usage; paid tier prices and numerical allowances are not listed in the launch announcement. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Settings, then Data controls, then Download your Muse data, covering chats, files and other agent information. No retention period after deletion is stated. [8]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Muse Secure VM with its own browser [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Built-in protected credential storage; the agent cannot see passwords. 1Password integration was announced as coming soon. [1]
- **Approval before actions** `documented` — Requests approval for emails and purchases; a separate Sentinel checks outgoing activity. [1]
- **What it remembers** `documented` — Learns from conversations and personal context [1]
- **Edit & export your data** `limited` (partial or conditional only) — Memory files are readable and editable; exporting the full runtime is not specified. [2]
- **Privacy & isolation** `documented` — User chooses access; dedicated VM [1]
- **Availability & access** `limited` (partial or conditional only) — Launched; current regional coverage needs review [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free | Everyday use through the Muse app and web | Numerical allowance not published; Meta also announces paid subscriptions for heavier use, without public tier prices | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web at muse.ai; no native desktop app is listed in the launch announcement. | 1 |
| iOS | Available | Native iOS and Android apps, plus WhatsApp access. | 1 |
| Android | Available | Native iOS and Android apps, plus WhatsApp access. | 1 |
| macOS | Available | Mac app announced as out on 17 September 2026, described as working across apps, files, calendar, notes and messages on the computer. | 7 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | Muse app and WhatsApp | 1 |

## Hardware

`announced` Announced, not shipping — **Meta AI glasses**: Meta says Muse is coming to its AI glasses in the coming months, where you say the agent's name and it can act on what you are looking at. Meta's own pages still describe this as coming soon, and the AI glasses release notes do not list it as shipped. Muse's cloud virtual machine is separate from this. [1, 9]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

| Data source | Open to third parties? | This agent's route | Detail | Sources |
| --- | --- | --- | --- | --- |
| Instagram & Facebook profile | First-party only | First-party | Reaches Instagram and Facebook profile and engagement data through Meta's Accounts Center. No third-party agent has an equivalent route to a personal Meta graph. | 4 |
| Instagram direct messages | Reachable by nobody | Cannot reach it | Meta states its generative AI excludes private messages, so Instagram DMs are out of reach for Meta's own assistant as well as for others. | 5 |
| WhatsApp messages | Reachable by nobody | Cannot reach it | WhatsApp personal messages are end-to-end encrypted and WhatsApp states they are not seen by Meta AI. | 6 |

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 60878 posts, 13490 positive and 4694 negative rated user-weeks — 74.2% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: App Store, Play Store, Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Getting tasks done | 3830 | 926 | 80.5% | https://feedbackbench.com/criteria/act.md |
| Talking to the agent | 1405 | 898 | 61% | https://feedbackbench.com/criteria/ask.md |
| Getting access and set up | 587 | 1270 | 31.6% | https://feedbackbench.com/criteria/access.md |
| Running on its own | 1151 | 264 | 81.3% | https://feedbackbench.com/criteria/auto.md |
| Paying for the agent | 878 | 456 | 65.8% | https://feedbackbench.com/criteria/billing.md |
| Connecting accounts, sites and devices | 615 | 622 | 49.7% | https://feedbackbench.com/criteria/connect.md |
| Trusting it with data and safety | 163 | 559 | 22.6% | https://feedbackbench.com/criteria/trust.md |
| Staying in control | 363 | 236 | 60.6% | https://feedbackbench.com/criteria/control.md |
| Dealing with the vendor | 77 | 423 | 15.4% | https://feedbackbench.com/criteria/vendor.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-17 · Release · [Muse for Mac released](https://x.com/finkd/status/2100713341555712149) · Meta — Meta's chief executive announces a Mac app nine days after launch, describing access to apps, files, calendar, notes and messages with user-controlled permissions.
- 2026-09-08 · Launch · [Meta introduces Muse](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/) · Meta — Meta announces a personal agent in the Muse app and WhatsApp, running on a dedicated virtual machine.

## Sources

1. [Meta introduces Muse, its plans and future glasses access](https://about.fb.com/news/2026/09/introducing-muse-personal-ai-agent/) — checked 2026-10-21
2. [How We Designed Muse](https://introducing.muse.ai/) — checked 2026-10-21
3. [Meta investor relations: stock listing](https://investor.atmeta.com/stock-info/default.aspx) — checked 2026-10-21
4. [Meta: building toward a smarter, more personalized assistant](https://about.fb.com/news/2025/01/building-toward-a-smarter-more-personalized-assistant/) — checked 2026-10-22
5. [Meta generative AI privacy guide](https://www.facebook.com/privacy/guide/genai/) — checked 2026-10-22
6. [WhatsApp privacy questions: Meta AI and personal messages](https://www.whatsapp.com/privacyquestions) — checked 2026-10-22
7. [Meta CEO announces Muse for Mac](https://x.com/finkd/status/2100713341555712149) — checked 2026-10-22
8. [Muse data controls and download](https://www.meta.com/help/artificial-intelligence/2225571704857152/) — checked 2026-10-22
9. [Introducing Ray-Ban Meta Audio and more AI glasses styles, plus Muse](https://about.fb.com/news/2026/09/introducing-ray-ban-meta-audio-glasses-new-styles-plus-muse/) — checked 2026-10-26

---

# nanobot

> A Python personal-agent framework with persistent workflows across terminal, browser, and chat apps.

- **Maker:** HKUDS
- **Catalogue type:** Self-hosted runtime
- **Website:** https://nanobot.wiki
- **Source code:** https://github.com/HKUDS/nanobot
- **Also known as:** unknown — no other names recorded
- **Evidence:** 9 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/nanobot/
- **This record:** https://agentlist.wiki/agents/nanobot.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1, 8 |
| Included in another plan | Codex, Grok, or Copilot account sign-in | documented | 1, 8 |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT | documented | 1 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | Research group | documented | 9 |
| Launch date | Feb 2, 2026 | documented | 9 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Research group [9]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 2, 2026 · Open-source release [9]

**Launch history**

- 2026-02-02 · Open-source release [9]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Dream long-term memory, long-horizon goals, and built-in Office/PDF file reading. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Browserbase integration through MCP; web search and page fetching are built in. [4, 5]
- **Voice & ambient listening** `documented` — Transcribes voice messages and WebUI microphone input into editable text. [4]
- **Background work & triggers** `documented` — Cron, scheduled automations, and long-horizon goals. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Browser interface and terminal on macOS, Windows, and Linux; native desktop app not specified. [1]
- **iPhone & Android apps** `limited` (partial or conditional only) — iOS home-screen web app (PWA); native iOS or Android app not specified. [7]
- **Web & messaging** `documented` — WebUI, Telegram, Discord, Slack, WeChat, email, and others. [1]
- **Messaging services** `documented` — Telegram, Slack, Discord, WeChat, Email [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Python SDK plus self-hosted OpenAI-compatible chat API with streaming and file uploads. [1, 6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Hosted providers, OpenAI-compatible endpoints, and local servers. [1]
- **Bring your own model** `documented` — Own provider keys or local servers including Ollama, vLLM, and LM Studio. [8]
- **Included in another plan** `documented` — Supports Codex, eligible X Premium/Grok, and GitHub Copilot account sign-in; nanobot supplies no model account. [1, 8]
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1, 8]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [1]
- **Where it runs** `documented` — Self-hosted local or server gateway. [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Can load secrets through 1Password, Bitwarden, or pass command-line wrappers; no built-in autofill specified. [4]
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Saved sessions and Dream long-term memory. [1]
- **Edit & export your data** `limited` (partial or conditional only) — Python SDK and OpenAI-compatible API. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — Documented channel allowlists and environment-based secret configuration. [2]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT open-source license; Own provider keys or local servers including Ollama, vLLM, and LM Studio. | Usage limits and running costs depend on your chosen provider and deployment. | 1, 8 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Browser interface and terminal on macOS, Windows, and Linux; native desktop app not specified. | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Telegram | Available | WebUI, Telegram, Discord, Slack, WeChat, email, and others. | 1 |
| Slack | Available | WebUI, Telegram, Discord, Slack, WeChat, email, and others. | 1 |
| Discord | Available | WebUI, Telegram, Discord, Slack, WeChat, email, and others. | 1 |
| WeChat | Available | WebUI, Telegram, Discord, Slack, WeChat, email, and others. | 1 |
| Email | Available | WebUI, Telegram, Discord, Slack, WeChat, email, and others. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-15 · Release · [nanobot v0.3.5 unifies terminal and browser continuity](https://github.com/HKUDS/nanobot/releases/tag/v0.3.5) — The release adds a native terminal client sharing saved conversations with the browser, plus improved automation views and session recovery.
- 2026-07-25 · Release · [nanobot v0.3.0](https://github.com/HKUDS/nanobot/releases/tag/v0.3.0) · HKUDS — 260 merged pull requests and 38 new contributors. A single command now prepares the local web interface, starts the gateway and opens a browser workbench, with guided provider setup on a fresh install.

## Sources

1. [nanobot README](https://raw.githubusercontent.com/HKUDS/nanobot/main/README.md) — checked 2026-10-21
2. [nanobot security policy](https://raw.githubusercontent.com/HKUDS/nanobot/main/SECURITY.md) — checked 2026-10-21
3. [nanobot v0.1.4.post6 README: OpenClaw inspiration](https://raw.githubusercontent.com/HKUDS/nanobot/v0.1.4.post6/README.md) — checked 2026-10-21
4. [nanobot configuration](https://github.com/HKUDS/nanobot/blob/main/docs/configuration.md) — checked 2026-10-21
5. [nanobot MCP presets](https://github.com/HKUDS/nanobot/blob/main/docs/mcp-presets.md) — checked 2026-10-21
6. [nanobot agent HTTP API](https://github.com/HKUDS/nanobot/blob/main/docs/openai-api.md) — checked 2026-10-21
7. [nanobot iOS home-screen web app](https://github.com/HKUDS/nanobot/blob/main/docs/guides/ios-pwa-status-bar.md) — checked 2026-10-21
8. [nanobot model providers](https://github.com/HKUDS/nanobot/blob/main/docs/providers.md) — checked 2026-10-21
9. [nanobot release archive: original launch](https://github.com/HKUDS/nanobot/blob/main/docs/release-archive.md) — checked 2026-10-21

---

# NanoClaw

> A customizable personal assistant running individual agents in isolated containers.

- **Maker:** NanoCo
- **Catalogue type:** Self-hosted runtime
- **Website:** https://nanoclaw.dev
- **Source code:** https://github.com/nanocoai/nanoclaw
- **Also known as:** unknown — no other names recorded
- **Evidence:** 7 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/nanoclaw/
- **This record:** https://agentlist.wiki/agents/nanoclaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1, 2 |
| Included in another plan | ChatGPT or Cursor accounts via optional providers | documented | 1 |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT | documented | 2 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 2 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | Startup | documented | 6, 7 |
| Launch date | Feb 2026 | documented | 6 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [6, 7]
- **Company stage** `documented` — Seed [6, 7]
- **Disclosed funding** `documented` — US$12M · Seed · May 20, 2026 [6, 7]
- **Launch date** `documented` — Feb 2026 · Open-source release [6]

**Launch history**

- 2026-02 · Open-source release [6]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Separate agent containers and reusable agent templates containing instructions, tools, and skills. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Bundled agent-browser skill opens pages, clicks, fills forms, and captures screenshots inside the agent container. [3]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Recurring tasks with optional script gates. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — Slack, Telegram, WhatsApp, Discord, iMessage, and installable adapters. [1]
- **Messaging services** `documented` — iMessage, WhatsApp, Telegram, Slack, Discord [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — MCP tools and a TypeScript channel-adapter interface; public REST API not specified. [1, 4]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Claude default; optional Codex, OpenCode, Cursor, or Ollama. [1]
- **Bring your own model** `documented` — Anthropic-compatible endpoints, optional Codex/OpenCode/Cursor providers, and local Ollama models. [1]
- **Included in another plan** `documented` — Optional Codex provider uses a ChatGPT subscription; optional Cursor provider uses a Cursor account. [1]
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1, 2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [2]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [2]
- **Where it runs** `documented` — Self-hosted Docker on macOS, Linux, or WSL2. [1]
- **Passwords & credentials** `documented` — Separate credential gateway, currently OneCLI Agent Vault; injects API credentials outside agent containers. [1]
- **Approval before actions** `documented` — Admin approval for package installs and MCP additions; the credential gateway can gate credentialed requests. [5]
- **What it remembers** `documented` — Separate per-agent workspace, instructions, and memory. [1]
- **Edit & export your data** `limited` (partial or conditional only) — User-customized code forks; reusable agent templates. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — Container filesystem isolation and credential gateway. [1]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT open-source license; Anthropic-compatible endpoints, optional Codex/OpenCode/Cursor providers, and local Ollama models. | Usage limits and running costs depend on your chosen provider and deployment. | 1, 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

unknown — no native app or web client is recorded.

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Slack, Telegram, WhatsApp, Discord, iMessage, and installable adapters. | 1 |
| WhatsApp | Available | Slack, Telegram, WhatsApp, Discord, iMessage, and installable adapters. | 1 |
| Telegram | Available | Slack, Telegram, WhatsApp, Discord, iMessage, and installable adapters. | 1 |
| Slack | Available | Slack, Telegram, WhatsApp, Discord, iMessage, and installable adapters. | 1 |
| Discord | Available | Slack, Telegram, WhatsApp, Discord, iMessage, and installable adapters. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-09 · Release · [NanoClaw 2026.10.0 moves to calendar versioning](https://github.com/nanocoai/nanoclaw/releases/tag/v2026.10.0) · NanoClaw — The first release numbered by calendar date and the first the updater installs by default, so updates follow published releases rather than the tip of the main branch, with more dependable updates and rollbacks.

## Sources

1. [NanoClaw README](https://raw.githubusercontent.com/nanocoai/nanoclaw/main/README.md) — checked 2026-10-21
2. [NanoClaw official repository and license](https://github.com/nanocoai/nanoclaw) — checked 2026-10-21
3. [NanoClaw bundled browser automation](https://github.com/nanocoai/nanoclaw/blob/main/container/skills/agent-browser/SKILL.md) — checked 2026-10-21
4. [NanoClaw channel adapter interface](https://github.com/nanocoai/nanoclaw/blob/main/docs/api-details.md) — checked 2026-10-21
5. [NanoClaw approval-gated tools](https://github.com/nanocoai/nanoclaw/blob/main/src/modules/approvals/agent.md) — checked 2026-10-21
6. [NanoCo seed financing and NanoClaw launch history](https://www.helpnetsecurity.com/2026/05/20/nanoco-seed-funding-12-million/) — checked 2026-10-21
7. [NanoCo announces its seed financing](https://www.linkedin.com/posts/nanocoai_today-were-announcing-nanoco-our-enterprise-activity-7462898172324143105-beHf) — checked 2026-10-21

---

# Noah

> An AI chief of staff reached by text and voice rather than an app, which schedules, chases follow-ups and surfaces insights. Pricing and sign-up are not published.

- **Maker:** Hey Noah AI Inc.
- **Catalogue type:** Under review
- **Website:** https://www.heynoah.io
- **Source code:** unknown — no public repository recorded
- **Also known as:** Hey Noah
- **Evidence:** 3 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/heynoah/
- **This record:** https://agentlist.wiki/agents/heynoah.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Access unverified · age limits | documented | 1, 3 |
| Price & usage costs | Not published | limited | 3 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Provider cloud | documented | 2 |
| Open source & license | unknown | unknown | — |
| Controls your computer | No | documented | 2 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 2 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 2 |
| Company type | Startup | documented | 1 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [1]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Presented as having its own identity and working across your network rather than waiting to be prompted, reached through voice and text instead of an app [1]
- **Computer use** `documented` — Explicitly ruled out: the company states it does not access your screen, keystrokes, passwords or device location, and acts only through connected integrations [2]
- **Controls your computer** `documented` — The company states it does not access your screen, keystrokes, passwords or device location. [2]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Not offered; the company states it does not access your screen or collect screenshots [2]
- **Voice & ambient listening** `documented` — Voice is a primary channel, alongside phone calls [1, 3]
- **Background work & triggers** `limited` (partial or conditional only) — Described as proactively helping rather than waiting for prompts, but no named scheduling or trigger feature is published [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — No desktop app; the product is reached through messaging and a web application [1]
- **iPhone & Android apps** `documented` — No mobile app; the product is reached through messaging [1]
- **Web & messaging** `documented` — Text messages, RCS, WhatsApp, iMessage, email, Slack and phone calls, plus a web application [2, 3]
- **Messaging services** `documented` — iMessage, RCS, SMS, WhatsApp, Slack, Email [2, 3]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — Third-party large language model providers, which are not named [2]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `limited` (partial or conditional only) — Not published. The terms state only that certain features may require payment and that fees may change on reasonable notice [3]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted with no component on the user's device. The company has not published what happens if it shuts down. [2]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A right to receive your data in a structured, commonly used format by writing to the company. [2]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Provider-hosted; no component runs on the user's device [2]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `limited` (partial or conditional only) — Keeps a copy of connected calendars to stay in sync and stores copies of the messages it works with; no user-facing memory view is described [2]
- **Edit & export your data** `documented` — A right to receive your data in a structured, commonly used format by writing to the company; no in-product export is described [2]
- **Privacy & isolation** `documented` — The company states it does not train models on customer data and does not collect screen captures or keystrokes [2]
- **Availability & access** `documented` — No self-serve sign-up is published: the site offers only a demo and a login, and users must be eighteen or older [1, 3]

## Plans

unknown — no published plan or price list is recorded. See the Price & usage costs claim above.

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web application at app.heynoah.io, reached by login. | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Listed in the privacy policy and terms as a channel. | 2, 3 |
| RCS | Available | Listed in the privacy policy and terms as a channel. | 2, 3 |
| SMS | Available | Listed in the privacy policy and terms as a channel. | 2, 3 |
| WhatsApp | Available | Listed in the privacy policy and terms as a channel. | 2, 3 |
| Slack | Available | Listed in the privacy policy and terms as a channel. | 2, 3 |
| Email | Available | Listed in the privacy policy and terms as a channel. | 2, 3 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Hey Noah official website](https://www.heynoah.io) — checked 2026-10-22
2. [Hey Noah privacy policy](https://www.heynoah.io/privacy) — checked 2026-10-22
3. [Hey Noah terms of service](https://www.heynoah.io/terms) — checked 2026-10-22

---

# Omi

> An always-listening personal AI that captures conversations and your screen, turning them into summaries, action items and searchable memory, with an optional pendant.

- **Maker:** Based Hardware Inc.
- **Catalogue type:** Local-first app
- **Website:** https://www.omi.me
- **Source code:** https://github.com/BasedHardware/omi
- **Also known as:** Friend, Based Hardware
- **Evidence:** 7 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/omi/
- **This record:** https://agentlist.wiki/agents/omi.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1, 4 |
| Price & usage costs | Free; Pro US$269/yr | documented | 4 |
| Included in another plan | Free plan included with every device; Pro is a standalone subscription | documented | 4 |
| Where it runs | Hosted, or self-host the backend | documented | 2 |
| Open source & license | MIT | documented | 3 |
| Controls your computer | Limited · Mac, Windows | documented | 1 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Partly | documented | 2, 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 5, 6 |
| Company type | Startup | documented | 1 |
| Launch date | Mar 22, 2024 | documented | 2 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [1]
- **Company stage** `documented` — Based Hardware Inc., San Francisco [1]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Mar 22, 2024 · Open-source release · Repository first published, under the earlier name Friend [2]

**Launch history**

- 2024-03-22 · Open-source release · Repository first published, under the earlier name Friend [2]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Captures conversations and screen context continuously and turns them into summaries, action items and searchable memory across phone, desktop and an optional pendant [1, 2]
- **Computer use** `limited` (partial or conditional only) — The desktop app reads the screen; driving a browser or desktop is not described [1]
- **Controls your computer** `documented` — The desktop app reads the screen for context but does not drive applications. [1]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Continuous listening and transcription is the core of the product, with on-phone transcription on the free plan [1, 4]
- **Background work & triggers** `limited` (partial or conditional only) — Capture is continuous and summarising automatic, but no scheduled or event-triggered autonomy is described for the user's own agent [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native macOS and Windows desktop apps [1]
- **iPhone & Android apps** `documented` — Native iOS and Android apps, plus an Apple Watch app [1, 7]
- **Web & messaging** `documented` — iPhone, Android, Apple Watch, macOS, Windows and the web [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `documented` — A US$179 pendant discounted to US$129 at checkout, with ten to fourteen hours of battery, shipping worldwide; a glasses developer kit is US$499. Neither is required, since the free plan works with phone-only transcription, and the hardware designs are open source. [1, 4]
- **Developer API** `documented` — A public developer API with bearer keys covering memories, conversations and action items, rate limited per key and per user, plus an MCP server [6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — The hosted backend names Soniox for speech recognition; the language model providers are not disclosed [2]
- **Bring your own model** `limited` (partial or conditional only) — Bringing your own key is possible only by self-hosting the backend; no local model mode is described for the consumer app [2]
- **Included in another plan** `documented` — Free plan included with every device; Pro is a standalone subscription [4]
- **Price & usage costs** `documented` — Free plan with unlimited on-phone transcription and 300 cloud minutes a month; Pro US$269 a year. The pendant is US$179 with a US$50 discount applied at checkout, and a glasses developer kit is US$499 [4]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — MIT-licensed with a self-hostable backend, so a technical owner could keep it running, though the consumer apps point at the hosted service by default. [2, 3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A copy of personal data can be requested by email; the developer API can also read your own conversations and memories. [5, 6]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [3]
- **Where it runs** `documented` — Apps on your own devices talking to a hosted backend, with the whole backend self-hostable from the repository [2]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Conversations, transcripts, summaries, action items and remembered facts, preferences and relationships persist and are searchable [1, 5]
- **Edit & export your data** `limited` (partial or conditional only) — A copy of personal data can be requested by email; no self-serve export is described, though the developer API can read your own conversations and memories [5, 6]
- **Privacy & isolation** `documented` — Deleting a conversation deletes its transcript and associated data including stored audio [5]
- **Availability & access** `documented` — Open sign-up with free apps on every platform; the device ships worldwide [1, 4]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Included with every device | Unlimited on-phone transcription; 300 cloud minutes a month | No subscription needed to start | 4 |
| Pro | US$269/year | Billed yearly | The full feature set | A monthly price is implied but not published | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web app at app.omi.me. | 1 |
| iOS | Available | Native iOS app on the App Store. | 7 |
| Android | Available | Native Android app on Google Play. | 1 |
| macOS | Available | Native macOS desktop app. | 1 |
| Windows | Available | Native Windows desktop app. | 1 |
| Apple Watch | Available | Apple Watch app. | 1 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`optional` Optional hardware — **Omi pendant**: A US$179 pendant discounted to US$129 at checkout, with ten to fourteen hours of battery, shipping worldwide; a glasses developer kit is US$499. Neither is required, since the free plan works with phone-only transcription, and the hardware designs are open source. [1, 4]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** no public release notes found — Checked: the blog carries auto-generated marketing content and no dated product updates (checked 2026-10-22)

No dated catalogue updates mention this agent yet.

## Sources

1. [Omi official website](https://www.omi.me) — checked 2026-10-22
2. [Omi repository](https://github.com/BasedHardware/omi) — checked 2026-10-22
3. [Omi licence](https://github.com/BasedHardware/omi/blob/main/LICENSE) — checked 2026-10-22
4. [Omi pricing](https://www.omi.me/pages/pricing) — checked 2026-10-22
5. [Omi privacy policy](https://www.omi.me/pages/privacy) — checked 2026-10-22
6. [Omi API reference](https://docs.omi.me/api-reference/introduction) — checked 2026-10-22
7. [Omi on the App Store](https://apps.apple.com/us/app/friend-ai-wearable/id6502156163) — checked 2026-10-22

---

# OpenAI Dots

> An always-on personal agent with its own cloud computer and context carried between conversations.

- **Maker:** OpenAI
- **Catalogue type:** Hosted agent
- **Website:** https://openai.com/index/introducing-dots/
- **Source code:** unknown — no public repository recorded
- **Also known as:** Dot, OpenAI Dot, ChatGPT dots
- **Evidence:** 8 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/dots/
- **This record:** https://agentlist.wiki/agents/dots.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · rolling out, paid plan, some regions | documented | 2 |
| Price & usage costs | ChatGPT Pro $100/mo | documented | 2, 4, 5 |
| Included in another plan | ChatGPT Pro / Business Premium | limited | 2 |
| Where it runs | Cloud + your PC | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | Limited | limited | 1 |
| Its own computer | Cloud | documented | 1 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 7, 8 |
| Company type | Major AI lab | documented | 6 |
| Launch date | Sep 29, 2026 | documented | 1 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Major AI lab [6]
- **Company stage** `documented` — Private [6]
- **Disclosed funding** `documented` — US$122B · Funding round · Mar 31, 2026 · Committed capital [6]
- **Launch date** `documented` — Sep 29, 2026 · Public launch [1]

**Launch history**

- 2026-09-29 · Public launch [1]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Shares ChatGPT context and connected apps; can delegate work into Codex cloud environments. [2, 3]
- **Computer use** `documented` — Own cloud computer, plus optional access to a connected personal computer. [1]
- **Controls your computer** `limited` (partial or conditional only) — An optional connection to a personal computer, alongside the cloud machine it normally works on. [1]
- **Its own computer** `documented` — Its own cloud computer. Whether you can watch or take over its screen is not specified. [1]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Operates its cloud browser; you can inspect the computer and take over. [1, 2]
- **Voice & ambient listening** `limited` (partial or conditional only) — Voice calls with your dot; it cannot initiate calls at launch. [1, 2]
- **Background work & triggers** `documented` — Ongoing and scheduled work [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Available in the ChatGPT desktop app, including Windows, and desktop web; supported OS details beyond Windows are not listed in the dots setup guide. [2]
- **iPhone & Android apps** `documented` — Available inside ChatGPT for iOS and Android; mobile web is unsupported. [2]
- **Web & messaging** `documented` — ChatGPT, Slack, Teams [1]
- **Messaging services** `documented` — Slack, Microsoft Teams [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — GPT-6 Astra at launch [1]
- **Bring your own model** `documented` — GPT-6 Astra is supplied by the service; BYO model keys are not listed for dots. [1]
- **Included in another plan** `limited` (partial or conditional only) — First dot included in ChatGPT Pro or Business Premium; Enterprise beta requires administrator enablement. [2]
- **Price & usage costs** `documented` — First dot included in ChatGPT Pro 100, 200 or 500 (US$100/200/500 per month) or Business Premium (US$125/user/month; US$100/user/month annually). Deeper-work allowances apply. [2, 4, 5]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — ChatGPT history and data export from the account's data controls. Business, Enterprise and Healthcare workspaces have no self-service export. Deleted data is removed within 30 days. [7, 8]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed cloud computer; optional connected devices [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Saved passwords and secure sign-in flows keep credentials out of the model; a standalone password vault is not specified. [1, 3]
- **Approval before actions** `documented` — Action review follows custom rules; purchases need approval and some authentication steps stay manual. [3]
- **What it remembers** `documented` — ChatGPT memory and ongoing context [2]
- **Edit & export your data** `limited` (partial or conditional only) — Deleting the dot clears its context; individual dot memories cannot currently be viewed or edited. [3]
- **Privacy & isolation** `documented` — Custom Rules and action review; proactive research is read-only [1]
- **Availability & access** `documented` — Gradual rollout; plan and regional restrictions [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| ChatGPT Pro 100 | US$100/month | Monthly | First dot; ChatGPT Pro tools and Codex | Pro rollout excludes EEA, Switzerland and UK; Deeper-work allowance; expanded launch limits | 2, 4 |
| ChatGPT Pro 200 | US$200/month | Monthly | First dot; more included Pro usage than Pro 100 | New subscriptions have a reduced allowance; Eligible existing allowance lasts through October 29, 2026 | 2, 4 |
| ChatGPT Pro 500 | US$500/month | Monthly | First dot; highest Pro allowance; Ultrafast service tier in ChatGPT Work and Codex | Ultrafast consumes the included allowance faster | 2, 4 |
| ChatGPT Business Premium | US$125/user/month | Monthly, or US$100/user/month billed annually | First dot and a managed workspace; Five times Standard seat usage | Subject to workspace controls; Deeper-work allowance; expanded launch limits | 2, 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Available in the ChatGPT desktop app, including Windows, and desktop web; supported OS details beyond Windows are not listed in the dots setup guide. | 2 |
| Windows | Available | Available in the ChatGPT desktop app, including Windows, and desktop web; supported OS details beyond Windows are not listed in the dots setup guide. | 2 |
| iOS | Available | Available inside ChatGPT for iOS and Android; mobile web is unsupported. | 2 |
| Android | Available | Available inside ChatGPT for iOS and Android; mobile web is unsupported. | 2 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Slack | Available | ChatGPT, Slack, Teams | 1 |
| Microsoft Teams | Available | ChatGPT, Slack, Teams | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 14581 posts, 1417 positive and 1877 negative rated user-weeks — 43% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Getting tasks done | 548 | 391 | 58.4% | https://feedbackbench.com/criteria/act.md |
| Talking to the agent | 232 | 502 | 31.6% | https://feedbackbench.com/criteria/ask.md |
| Running on its own | 447 | 253 | 63.9% | https://feedbackbench.com/criteria/auto.md |
| Paying for the agent | 143 | 339 | 29.7% | https://feedbackbench.com/criteria/billing.md |
| Connecting accounts, sites and devices | 108 | 254 | 29.8% | https://feedbackbench.com/criteria/connect.md |
| Staying in control | 96 | 220 | 30.4% | https://feedbackbench.com/criteria/control.md |
| Getting access and set up | 49 | 228 | 17.7% | https://feedbackbench.com/criteria/access.md |
| Dealing with the vendor | 13 | 192 | 6.3% | https://feedbackbench.com/criteria/vendor.md |
| Trusting it with data and safety | 14 | 98 | 12.5% | https://feedbackbench.com/criteria/trust.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-29 · Launch · [OpenAI introduces Dots](https://openai.com/index/introducing-dots/) · OpenAI — The launch introduces ongoing personal agents with cloud computers, connected apps, and user-defined permissions.

## Sources

1. [Introducing dots · OpenAI](https://openai.com/index/introducing-dots/) — checked 2026-10-21
2. [Dots eligible plans and launch limits](https://help.openai.com/en/articles/20001530-getting-started-with-your-dot) — checked 2026-10-21
3. [Dots privacy, security, and safety FAQs](https://help.openai.com/en/articles/20001529-dots-privacy-security-and-safety-faqs) — checked 2026-10-21
4. [ChatGPT Pro prices and allowances](https://help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers) — checked 2026-10-21
5. [ChatGPT Business Premium seat pricing](https://help.openai.com/en/articles/11391654-chatgpt-business-release-notes) — checked 2026-10-21
6. [OpenAI closes its March 2026 funding round](https://openai.com/index/accelerating-the-next-phase-ai/) — checked 2026-10-21
7. [OpenAI privacy policy](https://openai.com/policies/row-privacy-policy/) — checked 2026-10-22
8. [Exporting your ChatGPT data](https://help.openai.com/en/articles/7260999) — checked 2026-10-22

---

# OpenClaw

> A persistent assistant gateway connecting your devices, tools, models, and chat channels.

- **Maker:** OpenClaw Foundation
- **Catalogue type:** Self-hosted runtime
- **Website:** https://openclaw.ai
- **Source code:** https://github.com/openclaw/openclaw
- **Also known as:** unknown — no other names recorded
- **Evidence:** 15 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/openclaw/
- **This record:** https://agentlist.wiki/agents/openclaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1 |
| Included in another plan | No OpenClaw paid tier; provider accounts are separate | documented | 1 |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT | documented | 1 |
| Controls your computer | Yes · Mac | documented | 4 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | Partly | limited | 1 |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | Independent project | documented | 15 |
| Launch date | Nov 2025 | documented | 14 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Independent project [15]
- **Company stage** `documented` — Nonprofit foundation [15]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Nov 2025 · Open-source release · Originally Clawd [14]

**Launch history**

- 2025-11 · Open-source release · Originally Clawd [14]
- 2026-01-29 · Major update · Renamed OpenClaw [14]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — One persistent gateway connects messaging, device companions, and swappable model or coding-agent backends. [1, 8]
- **Computer use** `documented` — Controls local/paired Mac desktops and cloud desktops; Windows/Linux driver support is experimental. [4]
- **Controls your computer** `documented` — Drives a local or paired Mac desktop; Windows and Linux driver support is experimental. [4]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `limited` (partial or conditional only) — Depends on the deployment: a self-hosted install on your own computer stops with it, while a server install keeps running. [1]
- **Browser control** `documented` — Dedicated agent browser, existing signed-in Chrome, or a remote browser connection. [5]
- **Voice & ambient listening** `documented` — Continuous Talk conversations on macOS/iOS/Android and the web, with realtime voice options. [12]
- **Background work & triggers** `documented` — Persistent scheduler with channel and webhook delivery. [3]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — macOS menu-bar app, Windows Hub, and Linux companion apps. [8]
- **iPhone & Android apps** `documented` — Official iOS and Android companion apps; both connect to your running Gateway. [9, 10]
- **Web & messaging** `documented` — Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. [1]
- **Messaging services** `documented` — iMessage, WhatsApp, Telegram, Slack, Discord, Microsoft Teams [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Self-hosted OpenAI-compatible Chat Completions and Responses APIs; enable them in Gateway settings. [11]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Swappable hosted models, local models, and harness plugins. [1]
- **Bring your own model** `documented` — Swappable hosted/local model providers and agent harnesses, using your own credentials. [1]
- **Included in another plan** `documented` — OpenClaw has no paid tier; model and hosting accounts are separate. [1]
- **Price & usage costs** `documented` — No paid software tier; model and hosting costs depend on deployment. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [1]
- **Where it runs** `documented` — Self-hosted: local computer or server. [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Opt-in 1Password and HashiCorp Vault integrations; 1Password agent reads can reveal selected secrets to the model. [6, 7]
- **Approval before actions** `documented` — Configurable command approvals and allowlists; full-permission modes can bypass prompts. [13]
- **What it remembers** `documented` — Editable Markdown workspace files and searchable memory. [2]
- **Edit & export your data** `limited` (partial or conditional only) — State and credentials remain on operator hardware. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — Sender pairing; main-session tools use host unless sandboxing is configured. [1]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT open-source license; Swappable hosted/local model providers and agent harnesses, using your own credentials. | Usage limits and running costs depend on your chosen provider and deployment. | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | macOS menu-bar app, Windows Hub, and Linux companion apps. | 8 |
| Windows | Available | macOS menu-bar app, Windows Hub, and Linux companion apps. | 8 |
| Linux | Available | macOS menu-bar app, Windows Hub, and Linux companion apps. | 8 |
| iOS | Available | Official iOS and Android companion apps; both connect to your running Gateway. | 9, 10 |
| Android | Available | Official iOS and Android companion apps; both connect to your running Gateway. | 9, 10 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. | 1 |
| WhatsApp | Available | Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. | 1 |
| Telegram | Available | Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. | 1 |
| Slack | Available | Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. | 1 |
| Discord | Available | Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. | 1 |
| Microsoft Teams | Available | Telegram, WhatsApp, Discord, Slack, Teams, iMessage, and others. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

| Date | Thread | Platform | Kind | Sentiment | Summary |
| --- | --- | --- | --- | --- | --- |
| 2026-02-05 | [A week running OpenClaw on a Mac mini](https://news.ycombinator.com/item?id=46895546) | Hacker News | community | mixed | A business owner reports useful overnight research and integrations, alongside invented claims, brittle browser automation and continuing review work. |
| 2026-04-10 | [Users question OpenClaw's memory reliability](https://news.ycombinator.com/item?id=47721955) | Hacker News | community | critical | Participants describe inconsistent recurring workflows and update regressions, and discuss more deterministic tools and simpler agent setups. |
| 2026-04-15 | [Who is still using OpenClaw?](https://news.ycombinator.com/item?id=47783940) | Hacker News | community | mixed | Users share editable Obsidian memory, reminders and family-history workflows. Replies debate everyday usefulness, maintenance and consent. |

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 39195 posts, 3099 positive and 3808 negative rated user-weeks — 44.9% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: App Store, Play Store, Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Dealing with the vendor | 295 | 1071 | 21.6% | https://feedbackbench.com/criteria/vendor.md |
| Getting tasks done | 686 | 489 | 58.4% | https://feedbackbench.com/criteria/act.md |
| Running on its own | 744 | 409 | 64.5% | https://feedbackbench.com/criteria/auto.md |
| Getting access and set up | 201 | 756 | 21% | https://feedbackbench.com/criteria/access.md |
| Staying in control | 577 | 330 | 63.6% | https://feedbackbench.com/criteria/control.md |
| Talking to the agent | 311 | 349 | 47.1% | https://feedbackbench.com/criteria/ask.md |
| Paying for the agent | 174 | 441 | 28.3% | https://feedbackbench.com/criteria/billing.md |
| Connecting accounts, sites and devices | 274 | 239 | 53.4% | https://feedbackbench.com/criteria/connect.md |
| Trusting it with data and safety | 140 | 235 | 37.3% | https://feedbackbench.com/criteria/trust.md |

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-22 · Release · [OpenClaw 2026.10.6 requires a patched Node runtime](https://github.com/openclaw/openclaw/releases/tag/v2026.10.6) · OpenClaw Foundation — Now requires Node 24.22.1 or 26.12.1 and later, which reject TLS chains signed with the certificate keys compromised in the October 2026 Node.js security releases. Docker images and installers ship the patched runtime.
- 2026-10-18 · Release · [OpenClaw publishes v2026.10.5](https://github.com/openclaw/openclaw/releases/tag/v2026.10.5) — The GitHub release publishes the stable version and links its full changelog. The publication date differs from the version string.

## Ecosystem

- Built on: OpenClaw → Kimi Claw — Kimi documents deploying OpenClaw in its cloud and connecting existing OpenClaw installations. (https://www.kimi.com/en/help/kimi-claw/overview)
- Built on: OpenClaw → MaxClaw — MiniMax hosts OpenClaw as MaxClaw, with managed setup, model access, and a cloud workspace. (https://agent.minimax.io/activity/max-claw)

## Sources

1. [OpenClaw README](https://raw.githubusercontent.com/openclaw/openclaw/main/README.md) — checked 2026-10-21
2. [Memory overview](https://docs.openclaw.ai/concepts/memory) — checked 2026-10-21
3. [Automations](https://docs.openclaw.ai/automation/cron-jobs) — checked 2026-10-21
4. [OpenClaw computer use](https://github.com/openclaw/openclaw/blob/main/docs/nodes/computer-use.md) — checked 2026-10-21
5. [OpenClaw browser automation](https://github.com/openclaw/openclaw/blob/main/docs/tools/browser.md) — checked 2026-10-21
6. [OpenClaw 1Password plugin](https://github.com/openclaw/openclaw/blob/main/docs/plugins/onepassword.md) — checked 2026-10-21
7. [OpenClaw HashiCorp Vault integration](https://github.com/openclaw/openclaw/blob/main/docs/plugins/vault.md) — checked 2026-10-21
8. [OpenClaw platform apps](https://github.com/openclaw/openclaw/blob/main/docs/platforms/index.md) — checked 2026-10-21
9. [OpenClaw iOS app](https://github.com/openclaw/openclaw/blob/main/docs/platforms/ios.md) — checked 2026-10-21
10. [OpenClaw Android app](https://github.com/openclaw/openclaw/blob/main/docs/platforms/android.md) — checked 2026-10-21
11. [OpenClaw HTTP API](https://github.com/openclaw/openclaw/blob/main/docs/gateway/openai-http-api.md) — checked 2026-10-21
12. [OpenClaw Talk mode](https://github.com/openclaw/openclaw/blob/main/docs/nodes/talk.md) — checked 2026-10-21
13. [OpenClaw command approvals](https://github.com/openclaw/openclaw/blob/main/docs/tools/exec-approvals.md) — checked 2026-10-21
14. [OpenClaw: original launch and naming history](https://openclaw.ai/blog/introducing-openclaw) — checked 2026-10-21
15. [OpenClaw blog: formation of the independent foundation](https://openclaw.ai/blog) — checked 2026-10-21

---

# OpenFang

> A Rust agent operating system whose autonomous packages perform ongoing scheduled work.

- **Maker:** RightNow AI
- **Catalogue type:** Self-hosted runtime
- **Website:** https://openfang.sh
- **Source code:** https://github.com/RightNow-AI/openfang
- **Also known as:** unknown — no other names recorded
- **Evidence:** 5 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/openfang/
- **This record:** https://agentlist.wiki/agents/openfang.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1, 5 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT | documented | 1 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | unknown | unknown | — |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Prebuilt autonomous Hands for ongoing research, lead generation, browser tasks, and publishing workflows. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Playwright browser automation with persistent sessions for forms and multi-step workflows. [1]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Autonomous Hands and scheduled agent runs. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native Tauri apps for macOS, Windows, and Linux. [3]
- **iPhone & Android apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Web & messaging** `documented` — Messaging adapters including Telegram; browser dashboard. [1]
- **Messaging services** `documented` — Telegram [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — REST, WebSocket, and streaming APIs; also exposes an OpenAI-compatible chat endpoint. [1, 4]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Anthropic, Gemini, OpenAI-compatible providers, Ollama, vLLM. [1]
- **Bring your own model** `documented` — Own provider keys or local Ollama, vLLM, and LM Studio servers. [5]
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1, 5]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [1]
- **Where it runs** `documented` — Self-hosted daemon and dashboard. [1]
- **Passwords & credentials** `limited` (partial or conditional only) — Built-in encrypted credential vault for extensions; website-login autofill not specified. [1]
- **Approval before actions** `documented` — Browser Hand requires approval before purchases; Twitter Hand queues posts for approval. [1]
- **What it remembers** `documented` — SQLite persistence and vector embeddings. [1]
- **Edit & export your data** `limited` (partial or conditional only) — Imports OpenClaw agents, history, skills, and configuration. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — WASM sandbox, capability gates, audit chain; pre-1.0. [1]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT open-source license; Own provider keys or local Ollama, vLLM, and LM Studio servers. | Usage limits and running costs depend on your chosen provider and deployment. | 1, 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Messaging adapters including Telegram; browser dashboard. | 1 |
| macOS | Available | Native Tauri apps for macOS, Windows, and Linux. | 3 |
| Windows | Available | Native Tauri apps for macOS, Windows, and Linux. | 3 |
| Linux | Available | Native Tauri apps for macOS, Windows, and Linux. | 3 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Telegram | Available | Messaging adapters including Telegram; browser dashboard. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-05-12 · Release · [OpenFang v0.6.9 security patches](https://github.com/RightNow-AI/openfang/releases/tag/v0.6.9) · RightNow AI — Dependency security updates addressing advisories in certificate-chain parsing, including a reachable panic in revocation-list parsing and incorrectly accepted name constraints, plus a Wasmtime patch.
- 2026-04-19 · Release · [OpenFang v0.6.0 credits Hermes-inspired features](https://github.com/RightNow-AI/openfang/releases/tag/v0.6.0) — The release adds multi-destination cron delivery, skill configuration templates, and a shared slash-command registry.

## Sources

1. [OpenFang README](https://raw.githubusercontent.com/RightNow-AI/openfang/main/README.md) — checked 2026-10-21
2. [OpenFang v0.6.0: Hermes-inspired capabilities](https://github.com/RightNow-AI/openfang/releases/tag/v0.6.0) — checked 2026-10-21
3. [OpenFang desktop app](https://github.com/RightNow-AI/openfang/blob/main/docs/desktop.md) — checked 2026-10-21
4. [OpenFang API reference](https://github.com/RightNow-AI/openfang/blob/main/docs/api-reference.md) — checked 2026-10-21
5. [OpenFang LLM provider guide](https://github.com/RightNow-AI/openfang/blob/main/docs/providers.md) — checked 2026-10-21

---

# OpenHuman

> A local-first desktop personal AI that drives your real browser and desktop apps, with editable memory and more than a dozen messaging channels.

- **Maker:** Tiny Human Intelligence Inc
- **Catalogue type:** Self-hosted runtime
- **Website:** https://tinyhumans.ai/openhuman
- **Source code:** https://github.com/tinyhumansai/openhuman
- **Also known as:** Tiny Humans
- **Evidence:** 11 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/openhuman/
- **This record:** https://agentlist.wiki/agents/openhuman.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1, 2 |
| Price & usage costs | Free app; optional inference tiers | documented | 4 |
| Included in another plan | Free open-source app, with optional paid hosted inference | documented | 4 |
| Where it runs | Your desktop or your server | documented | 1, 2 |
| Open source & license | GPL-3.0 | documented | 3 |
| Controls your computer | Yes · Mac, Windows, Linux | documented | 6 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | Partly | limited | 2 |
| If the vendor stops | Keeps working | documented | 1, 3 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | limited | 11 |
| Company type | Startup | documented | 1 |
| Launch date | Feb 18, 2026 | documented | 2 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [1]
- **Company stage** `documented` — Tiny Human Intelligence Inc, Delaware [1]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 18, 2026 · Open-source release · Repository first published [2]

**Launch history**

- 2026-02-18 · Open-source release · Repository first published [2]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Drives your real Chrome and desktop apps while asking before anything it cannot undo, and is built to run many agents cheaply on modest hardware [2, 6]
- **Computer use** `documented` — Operates the user's own desktop applications, choosing targets from on-screen controls without screenshots, and stopping before any payment [6]
- **Controls your computer** `documented` — Drives the user's real Chrome and desktop applications, asking before anything irreversible. [6]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `limited` (partial or conditional only) — Runs on the user's desktop, though a Linux server can host the core headless so work continues without a desktop session. [2]
- **Browser control** `documented` — Drives a real Chrome browser on the user's machine [6]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — A native cron tool plus durable workflow graphs with cron, event or manual triggers, approval steps and resume after a pause [2, 8]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native installers for macOS, Windows and Linux, plus a terminal app [1, 2]
- **iPhone & Android apps** `limited` (partial or conditional only) — An iOS companion is documented as experimental and not shipping [10]
- **Web & messaging** `documented` — Desktop app and terminal, plus fourteen messaging channels [1, 7]
- **Messaging services** `documented` — iMessage, WhatsApp, Telegram, Signal, Slack, Discord, WeChat, Feishu, Email [7]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — A JSON-RPC core over loopback or remote with a bearer token, a typed Rust embedding API, and an opt-in MCP server [2]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Bring your own key across 26 providers, plus local models through Ollama, LM Studio and MLX, with automatic routing [9]
- **Bring your own model** `documented` — Bring your own keys, or run local models; the optional hosted tiers buy cheaper inference rather than features [4, 9]
- **Included in another plan** `documented` — Free open-source app, with optional paid hosted inference [4]
- **Price & usage costs** `documented` — The app is free. Optional hosted inference is Basic US$19.99/month or US$199/year and Pro US$199.99/month or US$1,799.99/year; every tier has every feature and higher tiers only buy cheaper inference [4]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — GPL-3.0 software running on your own machine; a vendor account is optional and only buys hosted inference. [1, 3]
- **Remote shut-off** `documented` — Runs locally with no activation requirement; the account is optional. [1]
- **What you keep** `limited` (partial or conditional only) — State lives in a local folder you can copy, but no export feature is described and portability is offered only on request. [11]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — GPL-3.0 [3]
- **Where it runs** `documented` — Runs on the user's own Mac, Windows or Linux desktop, and can run headless on a Linux server; an optional hosted backend provides inference and billing [1, 2]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Asks before anything it cannot undo and stops before any payment [6]
- **What it remembers** `documented` — Remembers what you are working on, what you prefer and what is in your documents; the interface lets you add, delete, forget an item or forget a whole source [5]
- **Edit & export your data** `limited` (partial or conditional only) — No export feature is described; moving to another computer is a manual copy of the local data folder [11]
- **Privacy & isolation** `documented` — Sandboxing through an operating-system jail or Docker; memory can be stored on a self-hosted database instead of the vendor's [2, 5]
- **Availability & access** `documented` — Free download for macOS, Windows and Linux, described as an early beta under active development [1, 2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free app and free tier | Every feature; Bring your own keys or run local models | Hosted inference is charged per call | 4 |
| Basic | US$19.99/month | US$199/year | Roughly 50% cheaper hosted inference per call | Buys cheaper inference, not extra features | 4 |
| Pro | US$199.99/month | US$1,799.99/year | Roughly 90% cheaper hosted inference per call | Buys cheaper inference, not extra features | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Native macOS desktop app. | 1 |
| Windows | Available | Native Windows desktop app. | 1 |
| Linux | Available | Native Linux desktop app. | 1 |
| iOS | Preview | An iOS companion is documented as experimental and not part of the shipped product. | 10 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Listed among OpenHuman's shipped channels. | 7 |
| WhatsApp | Available | Listed among OpenHuman's shipped channels. | 7 |
| Telegram | Available | Listed among OpenHuman's shipped channels. | 7 |
| Signal | Available | Listed among OpenHuman's shipped channels. | 7 |
| Slack | Available | Listed among OpenHuman's shipped channels. | 7 |
| Discord | Available | Listed among OpenHuman's shipped channels. | 7 |
| WeChat | Available | Listed among OpenHuman's shipped channels. | 7 |
| Feishu | Available | Listed among OpenHuman's shipped channels. | 7 |
| Email | Available | Listed among OpenHuman's shipped channels. | 7 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-09 · Release · [OpenHuman v0.64.15](https://github.com/tinyhumansai/openhuman/releases/tag/v0.64.15) · Tiny Human Intelligence — A reliability release for installed builds, with a hotfix to module admission so critical modules such as search and memory load correctly again.

## Sources

1. [OpenHuman official page](https://tinyhumans.ai/openhuman) — checked 2026-10-22
2. [OpenHuman repository](https://github.com/tinyhumansai/openhuman) — checked 2026-10-22
3. [OpenHuman licence](https://github.com/tinyhumansai/openhuman/blob/main/LICENSE) — checked 2026-10-22
4. [OpenHuman billing and usage](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/billing-and-usage.md) — checked 2026-10-22
5. [OpenHuman memory](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/memory.md) — checked 2026-10-22
6. [OpenHuman browser and computer control](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/native-tools/browser-and-computer.md) — checked 2026-10-22
7. [OpenHuman channels](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/channels.md) — checked 2026-10-22
8. [OpenHuman workflows](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/workflows.md) — checked 2026-10-22
9. [OpenHuman local and BYOK models](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/model-routing/local-and-byok-models.md) — checked 2026-10-22
10. [OpenHuman iOS companion](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/features/ios-companion.md) — checked 2026-10-22
11. [Moving OpenHuman to a new computer](https://github.com/tinyhumansai/openhuman/blob/main/docs/gitbooks/en/guides/move-to-new-pc.md) — checked 2026-10-22

---

# Pally

> A personal assistant in messaging that remembers people, handles bookings and calls, and can work with your Mac or other people’s Pallys.

- **Maker:** Pally Technologies
- **Catalogue type:** Hosted agent
- **Website:** https://pally.com/
- **Source code:** unknown — no public repository recorded
- **Also known as:** Pally AI
- **Evidence:** 8 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/pally/
- **This record:** https://agentlist.wiki/agents/pally.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Free; paid Cloud Mac | documented | 2 |
| Included in another plan | unknown | unknown | — |
| Where it runs | unknown | unknown | — |
| Open source & license | unknown | unknown | — |
| Controls your computer | Yes · Mac | documented | 4 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 7 |
| Company type | Startup | documented | 6 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [6]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Makes calls and handles bookings; Trusted Pals lets personal assistants coordinate with per-person sharing controls. [1]
- **Computer use** `documented` — Mac companion can read the screen, access granted folders and photos, control apps and run terminal commands. The Mac must be online. [4]
- **Controls your computer** `documented` — A Mac companion reads the screen, opens granted folders and photos, controls apps and runs terminal commands. [4]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Completes website bookings and purchases. [2]
- **Voice & ambient listening** `documented` — Call Pally or ask it to call businesses for bookings, quotes and follow-ups. [3]
- **Background work & triggers** `documented` — Scheduled and event-triggered automations, morning briefs, important-message alerts and reminders. [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Pally Mac companion requires macOS 13 or later. [4]
- **iPhone & Android apps** `documented` — Conversations use existing messaging apps; no separate Pally phone app. [1]
- **Web & messaging** `documented` — iMessage, RCS, WhatsApp and Telegram; SMS is a fallback. [1, 2]
- **Messaging services** `documented` — iMessage, RCS, SMS, WhatsApp, Telegram [1, 2]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — MCP server exposes message_pally and check_pally with OAuth 2.1; a general REST API is not described. [5]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Free with a replenishing usage allowance. Cloud Mac is a paid add-on. Invite-only VIP costs $1,000/month or $10,000/year. [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — The product lets you ask what it remembers, disconnect accounts, download your data and delete your account in-product. No format is stated, and no retention period is given for adult accounts. [7]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Checks before sends, bookings or purchases that were not already requested or approved; unrequested Mac changes need approval. [1, 4]
- **What it remembers** `documented` — Remembers people, plans and personal details; memory can be viewed and edited. [1]
- **Edit & export your data** `documented` — View and edit remembered information, revoke Mac permissions and delete the account. [1, 4]
- **Privacy & isolation** `documented` — Publisher states encryption in transit and at rest. [1]
- **Availability & access** `documented` — Free signup without a card, invitation or waitlist. [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | $0 | Ongoing free plan | All connections and automations; Calls, personal number and email; Web research, bookings, purchases and memory; Trusted Pals sharing | Usage allowance refills through the week; chat continues after the limit. | 2 |
| Cloud Mac | Price not published | Optional paid add-on | Cloud Mac | — | 2 |
| VIP | $1,000/month or $10,000/year | Monthly or annual subscription | Higher usage limits; Same AI models as Free | Invitation only. | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Mac companion for screen, app, file and terminal access; macOS 13+. | 4 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | iMessage, RCS, WhatsApp and Telegram; SMS is a fallback. | 1, 2 |
| RCS | Available | iMessage, RCS, WhatsApp and Telegram; SMS is a fallback. | 1, 2 |
| SMS | Available | iMessage, RCS, WhatsApp and Telegram; SMS is a fallback. | 1, 2 |
| WhatsApp | Available | iMessage, RCS, WhatsApp and Telegram; SMS is a fallback. | 1, 2 |
| Telegram | Available | iMessage, RCS, WhatsApp and Telegram; SMS is a fallback. | 1, 2 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Pally product overview](https://pally.com/) — checked 2026-10-21
2. [Pally pricing and free plan](https://pally.com/pricing) — checked 2026-10-21
3. [Pally official product facts](https://pally.com/ai) — checked 2026-10-21
4. [Your Mac, from anywhere](https://pally.com/recipes/your-mac-from-anywhere) — checked 2026-10-21
5. [Pally MCP server](https://pally.com/mcp) — checked 2026-10-21
6. [Pally company profile at Y Combinator](https://www.ycombinator.com/companies/pally) — checked 2026-10-21
7. [Pally privacy policy](https://pally.com/privacy) — checked 2026-10-22
8. [Pally terms of service](https://pally.com/terms) — checked 2026-10-22

---

# Perplexity Computer

> A cloud agent with a real filesystem and browser, a memory called Brain that models your projects and people, and an optional desktop app that controls your own machine.

- **Maker:** Perplexity AI, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://www.perplexity.ai/computer
- **Source code:** unknown — no public repository recorded
- **Also known as:** Perplexity Brain, Comet
- **Evidence:** 11 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/perplexity-computer/
- **This record:** https://agentlist.wiki/agents/perplexity-computer.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · paid plan | documented | 1, 3 |
| Price & usage costs | Pro US$20/mo; Max US$200/mo | documented | 7, 8 |
| Included in another plan | With a Perplexity plan | documented | 7 |
| Where it runs | Provider cloud, or your hardware | documented | 1, 4, 5 |
| Open source & license | unknown | unknown | — |
| Controls your computer | Yes · Mac, Windows, Linux | documented | 4 |
| Its own computer | Cloud | documented | 2, 5 |
| Works while your computer is off | Keeps working | documented | 1, 6 |
| If the vendor stops | Stops | limited | 1 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Self-serve | documented | 11 |
| Company type | Startup | documented | 2 |
| Launch date | Feb 25, 2026 | documented | 2 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [2]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 25, 2026 · Public launch · Perplexity Computer introduced [2]

**Launch history**

- 2026-02-25 · Public launch · Perplexity Computer introduced [2]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Brain builds a wiki-style memory of concepts, entities and workstreams from your sessions, connectors and files, which you can read, edit and correct [3]
- **Computer use** `documented` — Cloud tasks get a real filesystem and browser on Perplexity's isolated machine; the separate Personal Computer desktop app controls the user's own machine including native apps like Mail and Finder [2, 4]
- **Controls your computer** `documented` — The Personal Computer desktop app reaches local files and native applications including Mail, Finder and Slack. [4]
- **Its own computer** `documented` — A secure isolated cloud machine with a real filesystem and browser; Portable Computer can instead run it on your own GPU hardware. [2, 5]
- **Works while your computer is off** `documented` — Cloud tasks and schedules run in Perplexity's environment rather than on the user's device. [1, 6]
- **Browser control** `documented` — A real browser in its isolated cloud environment [2]
- **Voice & ambient listening** `documented` — Voice control is listed as part of the Personal Computer desktop app [4]
- **Background work & triggers** `documented` — Scheduled Tasks run hourly, daily, weekly, monthly or on a custom cadence, at most once an hour and up to fifteen active per conversation, managed in an automations dashboard [6]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Desktop apps for macOS 15 and later, Windows 10 and 11, and Ubuntu [4]
- **iPhone & Android apps** `documented` — Native iOS and Android apps [1]
- **Web & messaging** `documented` — Web, iOS, Android and desktop apps, plus Slack and email [1, 9, 10]
- **Messaging services** `documented` — Slack, Email [9, 10]
- **Native hardware** `documented` — Portable Computer runs the model and agent on your own hardware, requiring an NVIDIA DGX Spark or an RTX-class GPU with at least 24 GB of video memory. It is optional; the cloud service needs no hardware. [5]
- **Developer API** `limited` (partial or conditional only) — A developer API platform exists, but a public API for controlling the Computer agent is not described [1]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Multi-provider, using Claude for core reasoning at launch alongside Gemini, Grok and ChatGPT as sub-agents; Portable Computer runs local models on your own hardware [2, 5]
- **Bring your own model** `limited` (partial or conditional only) — Custom API credentials are for third-party services the agent calls, not for model providers [1]
- **Included in another plan** `documented` — Included with a paid Perplexity subscription [7]
- **Price & usage costs** `documented` — Requires a paid plan: Pro US$20/month or US$200/year, Max US$200/month or US$2,000/year including 10,000 Computer credits a month; Pro includes no Computer credits. Enterprise Pro starts at US$40/month per seat [7, 8]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted and tied to a Perplexity subscription. The company has not published what happens if it is discontinued. [1]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — An export button on the account details page. [11]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — A secure isolated environment in Perplexity's cloud by default; Portable Computer runs the model and agent on your own hardware, and Personal Computer runs on your desktop [1, 4, 5]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Brain stores concepts, entities and workstreams drawn from sessions, connectors, files and your corrections, excluding incognito sessions [3]
- **Edit & export your data** `documented` — A self-serve export from the account details page [11]
- **Privacy & isolation** `documented` — Cloud tasks run in a secure isolated environment, and incognito sessions are excluded from memory [1, 3]
- **Availability & access** `documented` — Available to subscribers on web, iOS and Android; Brain remains a research preview limited to Max and Enterprise Max subscribers [1, 3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Pro | US$20/month | US$200/year | Access to Computer on web, iOS and Android | No Computer credits included; Brain is not available on Pro | 7 |
| Max | US$200/month | US$2,000/year, annual on the web only | 10,000 Computer credits a month; Brain research preview | Brain remains a research preview | 3, 8 |
| Enterprise Pro | From US$40/user/month | From US$400/year per seat | Team administration | Priced per seat | 7 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Available on the web to active subscribers. | 1 |
| iOS | Available | Native iOS app. | 1 |
| Android | Available | Native Android app. | 1 |
| macOS | Available | Personal Computer desktop app for macOS 15 and later. | 4 |
| Windows | Available | Personal Computer desktop app for Windows 10 and 11. | 4 |
| Linux | Available | Personal Computer desktop app for Ubuntu and DGX OS. | 4 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Slack | Available | Perplexity is usable inside Slack. | 9 |
| Email | Available | Computer is reachable by email. | 10 |

## Hardware

`optional` Optional hardware — **GPU hardware for Portable Computer**: Portable Computer runs the model and agent on your own hardware, requiring an NVIDIA DGX Spark or an RTX-class GPU with at least 24 GB of video memory. It is optional; the cloud service needs no hardware. [5]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.perplexity.ai/changelog — Dated changelog (checked 2026-10-22)

- 2026-10-19 · Release · [Ultrafast Computer models, Decisions API and connector controls](https://www.perplexity.ai/changelog/ultrafast-computer-models-decisions-api-and-connector-controls) · Perplexity — Faster models in Computer, a Decisions API for scored classification, and role-based connector permissions for enterprise administrators.
- 2026-10-05 · Release · [Automations, inline visualizations and new models](https://www.perplexity.ai/changelog/automations-inline-visualizations-and-gpt-6-1-sol) · Perplexity — Adds recurring Automations, inline data visualisations, additional frontier models, video generation and new connectors.
- 2026-09-21 · Release · [Effort Mode, new models and a Skills Marketplace](https://www.perplexity.ai/changelog/effort-mode-gpt-6-astra-and-skills-marketplace) · Perplexity — Introduces an effort setting trading intelligence against cost, a skills marketplace, and Portable Computer with hybrid compute on Mac.

## Sources

1. [What is Perplexity Computer](https://www.perplexity.ai/help-center/en/articles/13837784-what-is-computer) — checked 2026-10-22
2. [Introducing Perplexity Computer](https://www.perplexity.ai/hub/blog/introducing-perplexity-computer) — checked 2026-10-22
3. [What is Brain](https://www.perplexity.ai/help-center/en/articles/19700001-what-is-brain) — checked 2026-10-22
4. [What is Personal Computer](https://www.perplexity.ai/help-center/en/articles/14659663-what-is-personal-computer) — checked 2026-10-22
5. [What is Portable Computer](https://www.perplexity.ai/help-center/en/articles/20260915-what-is-portable-computer) — checked 2026-10-22
6. [Perplexity Tasks](https://www.perplexity.ai/help-center/en/articles/11521526-perplexity-tasks) — checked 2026-10-22
7. [Perplexity subscription plans](https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you) — checked 2026-10-22
8. [Perplexity Max](https://www.perplexity.ai/help-center/en/articles/11680686-perplexity-max) — checked 2026-10-22
9. [Perplexity in Slack](https://www.perplexity.ai/help-center/en/articles/14016915-using-perplexity-in-slack) — checked 2026-10-22
10. [Computer in email](https://www.perplexity.ai/help-center/en/articles/20260715-computer-in-email) — checked 2026-10-22
11. [Exporting your Perplexity data](https://www.perplexity.ai/help-center/en/articles/11564568-gdpr-compliance-at-perplexity) — checked 2026-10-22

---

# Persona

> A personal agent for messages, calls and errands, with an optional voice-activated wristband.

- **Maker:** Iris Assistant, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://yourpersona.com/
- **Source code:** unknown — no public repository recorded
- **Also known as:** yourpersona, @yourpersona
- **Evidence:** 4 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/persona/
- **This record:** https://agentlist.wiki/agents/persona.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Free; $179 Band | documented | 2 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Provider cloud | documented | 3 |
| Open source & license | Proprietary | documented | 4 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 3 |
| Company type | Startup | documented | 4 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [4]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Handles calls, appointments and purchases from messaging or an optional Band. [2, 4]
- **Computer use** `limited` (partial or conditional only) — Browser-based tasks; general desktop control is not specified. [3]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Operates a browser inside authorized services to complete tasks. [3]
- **Voice & ambient listening** `documented` — Outbound calls; the announced Band uses press/wrist activation, not continuous listening. [2]
- **Background work & triggers** `documented` — Standing instructions support reminders, mail triage and tasks that continue in the background. [4]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `documented` — iPhone and Android apps are listed in the official FAQ. [2]
- **Web & messaging** `documented` — iMessage, WhatsApp and a web dashboard. [1, 2]
- **Messaging services** `documented` — iMessage, WhatsApp [1, 2]
- **Native hardware** `limited` (partial or conditional only) — Optional wristband with microphones and speaker. $179 early-bird; December 2026 shipping. Software works without it. [2]
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Software is free; optional Band costs $179 at the early-bird price. [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A portable copy on request through a legal address, not a self-serve export. No retention period after deletion is stated. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — Proprietary software under a limited personal license. [4]
- **Where it runs** `documented` — Managed service using cloud infrastructure and third-party AI providers. [3]
- **Passwords & credentials** `limited` (partial or conditional only) — Stores authorized passwords, tokens and browser sessions; a user-facing vault is not specified. [3]
- **Approval before actions** `limited` (partial or conditional only) — FAQ promises approval for spending and messages; terms also allow transactions under standing authority. [2, 4]
- **What it remembers** `documented` — Persistent context from conversations, connected accounts and earlier tasks. [3]
- **Edit & export your data** `documented` — Export and deletion controls for account content. [1]
- **Privacy & isolation** `documented` — Encrypted in transit and at rest; the company reports SOC 2 Type I certification. [1]
- **Availability & access** `documented` — Messaging service available; Band ships in December 2026. [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Persona | Free | No subscription | Messaging and task service | Public usage caps are not listed | 2 |
| Persona Band | $179 | One-time early-bird purchase | Band, black strap, charging dock and cable | Shipping December 2026 | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| iOS | Available | iPhone app listed in the Band FAQ. | 2 |
| Android | Available | Android app listed in the Band FAQ. | 2 |
| Web | Limited availability | Web account dashboard. | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | iMessage, WhatsApp and a web dashboard. | 1, 2 |
| WhatsApp | Available | iMessage, WhatsApp and a web dashboard. | 1, 2 |

## Hardware

`announced` Announced, not shipping — **Persona Band**: Optional wristband with microphones and speaker. $179 early-bird; December 2026 shipping. Software works without it. [2]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Persona personal intelligence](https://yourpersona.com/) — checked 2026-10-21
2. [Persona Band, pricing and FAQ](https://yourpersona.com/band) — checked 2026-10-21
3. [Persona privacy policy](https://yourpersona.com/legal/privacy) — checked 2026-10-21
4. [Persona terms of service](https://yourpersona.com/legal/terms) — checked 2026-10-21

---

# Pickle

> An AI character with its own browser, outbound phone calls, a credential Vault, and agent-to-agent conversations.

- **Maker:** Pickle, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://www.pickle.com/
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 11 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/pickle/
- **This record:** https://agentlist.wiki/agents/pickle.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · age limits | documented | 2, 3 |
| Price & usage costs | From $20/mo | documented | 4 |
| Included in another plan | Standalone Pickle subscription. | documented | 4 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | Cloud | documented | 9 |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | documented | 9 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 7 |
| Company type | Startup | documented | 11 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [11]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — AI character that calls businesses, uses websites and coordinates with friends’ Pickles. [3]
- **Computer use** `limited` (partial or conditional only) — Uses its own computer for tasks; control of your local desktop is not specified. [9]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `documented` — A computer of its own, with isolated task execution. Whether you can watch it is not specified. [9]
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Navigates websites, completes forms and checkout; you can watch or take over. [2]
- **Voice & ambient listening** `documented` — In-app voice conversations and outbound calls to businesses. [2]
- **Background work & triggers** `documented` — One-time and repeating daily, weekly or monthly tasks. [6]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `documented` — iPhone app available; Android marked coming soon. [1, 2]
- **Web & messaging** `documented` — Text and voice within the Pickle app; phone calls. [2]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `limited` (partial or conditional only) — Previously announced glasses. An earlier official page says batch-one refunds are being processed; the current site focuses on the app. [1, 10]
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `limited` (partial or conditional only) — AI processing comes with a usage plan; personal model keys are not specified. [4, 8]
- **Included in another plan** `documented` — Standalone Pickle subscription. [4]
- **Price & usage costs** `documented` — Plus $20, Pro $50, Max $100, Ultra $200 per month; weekly credit allowances. [4]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Alone among hosted products here, the terms address discontinuation: where reasonable the company will say why and give users a chance to export their content. [9]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A copy of personal information on request. Account deletion starts immediately; diagnostic records go within 30 days, server logs within 90, and credential-use records are held in write-once storage until 365 days old. [7]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed service with isolated task execution. [1]
- **Passwords & credentials** `documented` — Built-in Vault keeps passwords and card details encrypted separately from the model. [7]
- **Approval before actions** `documented` — Requests approval for purchases, bookings and irreversible actions. [2]
- **What it remembers** `documented` — Remembers conversations; memories can be read, edited and deleted in Bubbles. [3]
- **Edit & export your data** `limited` (partial or conditional only) — Memory editing and deletion; full state export is not specified. [3]
- **Privacy & isolation** `documented` — Secrets bypass the model. Authorized staff may review conversations; the policy excludes advertising and model training. [8]
- **Availability & access** `documented` — iPhone app; 18 and older. [2, 3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Plus | $20/month | Monthly; renews automatically | 470 usage credits each week; Chat, calls and tasks | Resets Monday, 4 AM Pacific; Scheduled tasks pause when credits run out | 4, 5 |
| Pro | $50/month | Monthly; renews automatically | 1,170 usage credits each week; Chat, calls and tasks | Resets Monday, 4 AM Pacific; Scheduled tasks pause when credits run out | 4, 5 |
| Max | $100/month | Monthly; renews automatically | 2,330 usage credits each week; Chat, calls and tasks | Resets Monday, 4 AM Pacific; Scheduled tasks pause when credits run out | 4, 5 |
| Ultra | $200/month | Monthly; renews automatically | 4,670 usage credits each week; Chat, calls and tasks | Resets Monday, 4 AM Pacific; Scheduled tasks pause when credits run out | 4, 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| iOS | Available | Pickle iPhone app. | 2 |
| Android | Coming soon | Android is marked coming soon on the English homepage. | 1 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` Not established — **Pickle 1 glasses**: Previously announced glasses. An earlier official page says batch-one refunds are being processed; the current site focuses on the app. [1, 10]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** no public release notes found — Checked: the site is client-rendered and serves no readable changelog at any path (checked 2026-10-22)

No dated catalogue updates mention this agent yet.

## Sources

1. [Pickle personal agent](https://www.pickle.com/) — checked 2026-10-21
2. [Pickle iPhone app](https://apps.apple.com/us/app/pickle-your-personal-ai-agent/id6795790129) — checked 2026-10-21
3. [What is Pickle?](https://www.pickle.com/help/start/what-is-pickle) — checked 2026-10-21
4. [Pickle plans and weekly allowances](https://www.pickle.com/help/credits/plans) — checked 2026-10-21
5. [Pickle usage credits and limits](https://www.pickle.com/help/credits/usage-credits) — checked 2026-10-21
6. [Pickle scheduled tasks](https://www.pickle.com/help/tasks/scheduled) — checked 2026-10-21
7. [How Pickle protects your privacy](https://pickle.com/how-your-pickle-protects-your-privacy) — checked 2026-10-21
8. [Who can see your information?](https://www.pickle.com/help/privacy/who-can-see) — checked 2026-10-21
9. [Pickle terms of service](https://www.pickle.com/terms-of-service) — checked 2026-10-21
10. [Pickle 1 hardware announcement and refund notice](https://www.pickle.com/?pubDate=20260102) — checked 2026-10-21
11. [Pickle company profile at Y Combinator](https://www.ycombinator.com/companies/pickle-2) — checked 2026-10-21

---

# PicoClaw

> A Go personal assistant designed for small devices and diverse hardware.

- **Maker:** Sipeed
- **Catalogue type:** Self-hosted runtime
- **Website:** https://picoclaw.io
- **Source code:** https://github.com/sipeed/picoclaw
- **Also known as:** unknown — no other names recorded
- **Evidence:** 5 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/picoclaw/
- **This record:** https://agentlist.wiki/agents/picoclaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT | documented | 1 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | unknown | unknown | — |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Lightweight Go runtime for small Linux boards, RISC-V devices, and older Android phones. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Browser automation through the bundled agent-browser skill. [2]
- **Voice & ambient listening** `documented` — Speech-to-text and text-to-speech using configured providers; ambient listening not specified. [4, 5]
- **Background work & triggers** `documented` — One-shot reminders, recurring tasks, and cron. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Launcher opens chat in your browser; Windows/Linux system-tray support. Not a separate native chat UI. [1]
- **iPhone & Android apps** `documented` — Android APK runs PicoClaw on the phone; Termux is optional. [1]
- **Web & messaging** `documented` — Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. [1]
- **Messaging services** `documented` — WhatsApp, Telegram, Slack, Discord, WeChat, Matrix [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — MCP client for external tools; a developer HTTP API is not specified. [1]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Hosted providers plus Ollama and vLLM. [1]
- **Bring your own model** `documented` — Own API keys for hosted providers, or local Ollama/vLLM servers. [1]
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT [1]
- **Where it runs** `documented` — Self-hosted binary, Docker, or Android. [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `limited` (partial or conditional only) — Extensible tool-approval hooks can allow or deny actions; setup is required. [3]
- **What it remembers** `documented` — JSONL memory store. [1]
- **Edit & export your data** `limited` (partial or conditional only) — Binaries for RISC-V, ARM, MIPS, and x86. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — Pre-1.0 security caveats; configurable isolation documented. [1]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT open-source license; Own API keys for hosted providers, or local Ollama/vLLM servers. | Usage limits and running costs depend on your chosen provider and deployment. | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Launcher opens chat in your browser; Windows/Linux system-tray support. Not a separate native chat UI. | 1 |
| Android | Available | Android APK runs PicoClaw on the phone; Termux is optional. | 1 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. | 1 |
| Telegram | Available | Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. | 1 |
| Slack | Available | Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. | 1 |
| Discord | Available | Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. | 1 |
| WeChat | Available | Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. | 1 |
| Matrix | Available | Telegram, Discord, WhatsApp, Weixin, Slack, Matrix, and others. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-07-03 · Release · [PicoClaw v0.3.1](https://github.com/sipeed/picoclaw/releases/tag/v0.3.1) · Sipeed — A maintenance release collecting provider and storage fixes; the published notes list merged changes without a written summary.

## Sources

1. [PicoClaw README](https://raw.githubusercontent.com/sipeed/picoclaw/main/README.md) — checked 2026-10-21
2. [PicoClaw bundled browser skill](https://github.com/sipeed/picoclaw/blob/main/workspace/skills/agent-browser/SKILL.md) — checked 2026-10-21
3. [PicoClaw hook system](https://github.com/sipeed/picoclaw/blob/main/docs/architecture/hooks/README.md) — checked 2026-10-21
4. [PicoClaw speech recognition](https://github.com/sipeed/picoclaw/blob/main/pkg/audio/asr/README.md) — checked 2026-10-21
5. [PicoClaw speech synthesis](https://github.com/sipeed/picoclaw/blob/main/pkg/audio/tts/README.md) — checked 2026-10-21

---

# Poke

> A proactive assistant reached through familiar messaging apps, with integrations and shareable automation recipes.

- **Maker:** The Interaction Company
- **Catalogue type:** Hosted agent
- **Website:** https://poke.com
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 10 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/poke/
- **This record:** https://agentlist.wiki/agents/poke.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; $19 or $199/mo | documented | 2 |
| Included in another plan | Standalone Poke Pro or Ultra subscription | documented | 2 |
| Where it runs | Provider cloud | documented | 1 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | documented | 10 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 9 |
| Company type | Startup | documented | 7, 8 |
| Launch date | Sep 8, 2025 | documented | 7 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup · part of Cognition [7, 8]
- **Company stage** `documented` — Acquired · July 2026 [7, 8]
- **Disclosed funding** `documented` — US$15M · Seed · Sep 8, 2025 · Before joining Cognition [7]
- **Launch date** `documented` — Sep 8, 2025 · Public launch [7]

**Launch history**

- 2025-09-08 · Public launch [7]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Text-first assistant that can publish a website; Ultra can dispatch real-world tasks to human operators [2, 3]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Web search is included; interactive browser control is not specified [3]
- **Voice & ambient listening** `limited` (partial or conditional only) — Accepts voice messages; ambient listening is not specified [1]
- **Background work & triggers** `documented` — Reminders and scheduled tasks [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Messaging and web settings; native desktop app not specified [3]
- **iPhone & Android apps** `limited` (partial or conditional only) — Apple Messages, WhatsApp, Telegram and RCS; separate native app not specified [3]
- **Web & messaging** `documented` — Messages, WhatsApp, Telegram [1]
- **Messaging services** `documented` — iMessage, WhatsApp, Telegram [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Developer API sends messages into your agent; custom integrations can connect through MCP [4, 5]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `documented` — Standalone Poke Pro or Ultra subscription [2]
- **Price & usage costs** `documented` — Many features free; Pro $19/month; Ultra $199/month, with pay-as-you-go usage on top [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — The terms reserve the right to "change or stop providing the Service" with or without prior notice, and set out no wind-down or data-rescue provision. [10]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A portability right to a structured, machine-readable copy on request, offered in the EEA, UK and Switzerland section only. No self-serve export and no stated retention period after deletion. [9]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Hosted messaging service [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `limited` (partial or conditional only) — Personal context; memory implementation not verified [1]
- **Edit & export your data** `limited` (partial or conditional only) — Shareable Recipes; full memory export not verified [1]
- **Privacy & isolation** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Availability & access** `documented` — Sign-up available on official site [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free | Email, calendar and app connections | Lower usage limits than paid plans | 6 |
| Pro | US$19/month | Monthly | Frontier models for complex tasks; Background automations and priority support | Higher rate limits; numerical allowance not published | 6 |
| Ultra | US$199/month | Monthly | Frontier models for every action; Highest automation limits and support priority; Human operators for real-world tasks | Additional usage beyond included credits is paid separately; Purchases made by human operators are charged at cost from your credit balance | 2, 6 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Limited availability | Messaging and web settings; native desktop app not specified | 3 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Messages, WhatsApp, Telegram | 1 |
| WhatsApp | Available | Messages, WhatsApp, Telegram | 1 |
| Telegram | Available | Messages, WhatsApp, Telegram | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

| Date | Thread | Platform | Kind | Sentiment | Summary |
| --- | --- | --- | --- | --- | --- |
| 2026-05-01 | [Building an accountability bot with Poke recipes](https://www.reddit.com/r/buildinpublic/comments/1t19t5h/i_made_a_textbased_boss_bot_for_neurodivergent/) | Reddit | announcement | positive | A recipe creator shares scheduled check-ins, work timers and daily recaps built in Poke. A later reply asks whether the reminders sustain engagement. |

## Feedback snapshot

Provider: Feedback Bench (https://feedbackbench.com/); data: https://feedbackbench.com/data/personal.json. Window 2026-08-31 to 2026-10-18; checked 2026-10-21.
Sample: 509 posts, 73 positive and 84 negative rated user-weeks — 46.5% positive.
That percentage is the raw positive share of rated user-weeks, not the provider's composite score or ranking.
Channels: Reddit, X, YouTube. Build 1.1, built 2026-10-19.

| Area | Praise | Complaint | Positive share | Area method |
| --- | --- | --- | --- | --- |
| Dealing with the vendor | 1 | 30 | 3.2% | https://feedbackbench.com/criteria/vendor.md |
| Getting tasks done | 12 | 18 | 40% | https://feedbackbench.com/criteria/act.md |
| Talking to the agent | 19 | 10 | 65.5% | https://feedbackbench.com/criteria/ask.md |
| Paying for the agent | 6 | 9 | 40% | https://feedbackbench.com/criteria/billing.md |
| Running on its own | 6 | 9 | 40% | https://feedbackbench.com/criteria/auto.md |
| Trusting it with data and safety | 6 | 2 | 75% | https://feedbackbench.com/criteria/trust.md |
| Getting access and set up | 3 | 3 | 50% | https://feedbackbench.com/criteria/access.md |
| Staying in control | 2 | 4 | 33.3% | https://feedbackbench.com/criteria/control.md |
| Connecting accounts, sites and devices | 2 | 3 | 40% | https://feedbackbench.com/criteria/connect.md |

## Updates

**Where updates come from:** no public release notes found — Checked: no blog, changelog or updates page on the official site (checked 2026-10-22)

No dated catalogue updates mention this agent yet.

## Sources

1. [Poke · Official product page](https://poke.com/) — checked 2026-10-21
2. [Poke billing and human operators](https://poke.com/faq) — checked 2026-10-21
3. [Poke welcome and supported messaging channels](https://poke.com/docs) — checked 2026-10-21
4. [Poke developer API](https://poke.com/docs/api) — checked 2026-10-21
5. [Poke MCP servers](https://poke.com/docs/mcp-servers) — checked 2026-10-21
6. [Poke plans and usage](https://poke.com/pricing) — checked 2026-10-21
7. [Interaction announces Poke launch and seed funding](https://www.webwire.com/ViewPressRel.asp?aId=343533) — checked 2026-10-21
8. [Poke joins Cognition](https://poke.com/docs/release-notes) — checked 2026-10-21
9. [Poke privacy policy](https://poke.com/privacy) — checked 2026-10-22
10. [Poke terms of service](https://poke.com/terms) — checked 2026-10-22

---

# QwenPaw

> A self-hostable personal agent workstation with three layers of memory kept as editable Markdown, heartbeat check-ins and more than a dozen chat channels.

- **Maker:** AgentScope
- **Catalogue type:** Self-hosted runtime
- **Website:** https://qwenpaw.agentscope.io
- **Source code:** https://github.com/agentscope-ai/QwenPaw
- **Also known as:** AgentScope QwenPaw
- **Evidence:** 6 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/qwenpaw/
- **This record:** https://agentlist.wiki/agents/qwenpaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Free; free hosted option | documented | 2 |
| Included in another plan | Open-source software; the AgentScope Platform offers free hosted deployment | documented | 2 |
| Where it runs | Self-hosted or free cloud | documented | 2 |
| Open source & license | Apache-2.0 | documented | 3 |
| Controls your computer | Limited · Windows, Mac | documented | 2 |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 2, 3 |
| Remote shut-off | Not possible | documented | 2 |
| What you keep | Self-serve | documented | 2, 4 |
| Company type | Research group | documented | 2 |
| Launch date | Feb 27, 2026 | documented | 6 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Research group [2]
- **Company stage** `documented` — Built by the AgentScope team [2]
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `documented` — Feb 27, 2026 · Open-source release · First public release [6]

**Launch history**

- 2026-02-27 · Open-source release · First public release [6]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Memory is a personal knowledge base of structured Markdown files that are readable, editable, searchable and linked, alongside a MEMORY.md the user and agent both maintain [4]
- **Computer use** `limited` (partial or conditional only) — An optional Computer Use plugin operates approved desktop applications on the user's own machine, limited to Windows 10 and later or macOS 14 and later in the desktop app, and marked beta [2]
- **Controls your computer** `documented` — An optional Computer Use plugin operates approved desktop applications, in the desktop app only, marked beta and unavailable on Linux. [2]
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Launches its own Chromium, or drives the Chrome you are already signed in to through an extension [2]
- **Voice & ambient listening** `documented` — Voice and SIP are listed among its channels [5]
- **Background work & triggers** `documented` — Heartbeat runs at an interval and replies in the channel where you last chatted, alongside cron scheduled tasks that can run several recurring jobs in parallel [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — A Tauri desktop app for Windows 10 and later and macOS 14 and later, marked beta and unnotarised on macOS, plus a console web interface and a full terminal interface [2]
- **iPhone & Android apps** `documented` — No mobile app [2]
- **Web & messaging** `documented` — Desktop app, console web interface and terminal, plus more than a dozen chat channels [2, 5]
- **Messaging services** `documented` — iMessage, Telegram, Slack, Discord, WeChat, WeCom, Feishu, Matrix, Email [5]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — A REST API built on the AgentScope Runtime protocol, with an agent identifier header, plus ACP integration [2]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Built-in local QwenPaw-Flash models from 2B to 35B through llama.cpp, Ollama and LM Studio, plus keys for more than fourteen cloud providers including Qwen, OpenAI, Anthropic, Gemini, DeepSeek and Kimi [2]
- **Bring your own model** `documented` — Bring your own keys, or run the bundled local models [2]
- **Included in another plan** `documented` — Open-source software; the AgentScope Platform offers free hosted deployment [2]
- **Price & usage costs** `documented` — Free software. A hosted deployment on the AgentScope Platform is offered free; a cloud model key is required unless you run local models [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Apache-2.0 software you deploy yourself, so it keeps running if the project stops. [2, 3]
- **Remote shut-off** `documented` — Self-hosted with no activation or licence check described. [2]
- **What you keep** `documented` — A backup tool in settings exports the entire agent environment as a zip, and memory is plain Markdown on your own disk. [2, 4]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — Apache-2.0 [3]
- **Where it runs** `documented` — Self-hosted by pip, a script, Docker or the desktop app, or deployed to Alibaba Cloud, the AgentScope Platform or ModelScope [2]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Three layers: a live working context, the full verbatim history, and a personal knowledge base of structured Markdown in a local workspace directory [4]
- **Edit & export your data** `documented` — Settings include a backup tool that exports the whole agent environment as a zip containing workspaces, persona, memory, skills, chat history, settings and secrets [2]
- **Privacy & isolation** `documented` — Deploy locally and data stays on your machine; secrets are held in the agent environment [2]
- **Availability & access** `documented` — Open source; install by pip, script, Docker or the desktop app, with a free hosted option [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | US$0 | Free software | Full features when self-hosted; Bundled local models, or your own provider keys; Free hosted deployment on the AgentScope Platform | A cloud model key is required unless local models are used | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Console web interface served by your own instance. | 2 |
| Windows | Beta | Tauri desktop app for Windows 10 and later, marked beta. | 2 |
| macOS | Beta | Tauri desktop app for macOS 14 and later, marked beta and unnotarised. | 2 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Listed among QwenPaw's channels. | 5 |
| Telegram | Available | Listed among QwenPaw's channels. | 5 |
| Slack | Available | Listed among QwenPaw's channels. | 5 |
| Discord | Available | Listed among QwenPaw's channels. | 5 |
| WeChat | Available | Listed among QwenPaw's channels. | 5 |
| WeCom | Available | Listed among QwenPaw's channels. | 5 |
| Feishu | Available | Listed among QwenPaw's channels. | 5 |
| Matrix | Available | Listed among QwenPaw's channels. | 5 |
| Email | Available | Listed among QwenPaw's channels. | 5 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-22 · Release · [QwenPaw v2.2.2.post1](https://github.com/agentscope-ai/QwenPaw/releases/tag/v2.2.2.post1) · AgentScope — Security remediation removing emSign certificate authority trust across dependencies, installers and builds, and updating the bundled Node.js runtime.

## Sources

1. [QwenPaw official site](https://qwenpaw.agentscope.io) — checked 2026-10-22
2. [QwenPaw repository](https://github.com/agentscope-ai/QwenPaw) — checked 2026-10-22
3. [QwenPaw licence](https://github.com/agentscope-ai/QwenPaw/blob/main/LICENSE) — checked 2026-10-22
4. [QwenPaw memory documentation](https://github.com/agentscope-ai/QwenPaw/blob/main/website/public/docs/memory.en.md) — checked 2026-10-22
5. [QwenPaw channels documentation](https://github.com/agentscope-ai/QwenPaw/blob/main/website/public/docs/channels.en.md) — checked 2026-10-22
6. [QwenPaw first release](https://github.com/agentscope-ai/QwenPaw/releases) — checked 2026-10-22

---

# Rene

> A messaging assistant that coordinates email, calendars and browser errands, with Rene-to-Rene planning between people.

- **Maker:** Marumaru / Second Enlightenment
- **Catalogue type:** Hosted agent
- **Website:** https://rene.co/
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 6 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/rene/
- **This record:** https://agentlist.wiki/agents/rene.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2, 5 |
| Price & usage costs | Free to start | documented | 1, 5 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Provider cloud | documented | 3, 4 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 3, 4 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 3 |
| Company type | Startup | documented | 1 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [1]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Rene-to-Rene planning, connected-app coordination, and creating websites, slides and images from chat. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Browses websites and completes errands such as restaurant reservations. [1]
- **Voice & ambient listening** `documented` — Accepts voice requests. [6]
- **Background work & triggers** `documented` — Recurring briefs, reminders and follow-ups; surfaces changes requiring attention. [5]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `documented` — Native iPhone and iPad app; requires iOS / iPadOS 16.4 or later. [5]
- **Web & messaging** `documented` — Messages on iPhone, WhatsApp, Telegram and the web app. [2, 3]
- **Messaging services** `documented` — iMessage, WhatsApp, Telegram [2, 3]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — OpenAI and DeepSeek inference, with OpenRouter for routing. [4]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Free to start. US App Store lists Rene Pro purchases at $20 and $200, and Premium at $200; purchase periods and allowances are not listed. [1, 5]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [3, 4]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Access, correction, export or deletion by contacting the company; no self-serve export, no format stated and no retention period after cancellation. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed service with external AI inference providers. [3, 4]
- **Passwords & credentials** `limited` (partial or conditional only) — Email and calendar use delegated OAuth access rather than storing service passwords. [3]
- **Approval before actions** `documented` — Asks before sensitive information is sent or changed. [2]
- **What it remembers** `documented` — Keeps conversations, preferences, saved context and connected-app state for future requests. [3]
- **Edit & export your data** `documented` — Access, correction, export and deletion can be requested; connected services can be disconnected. [3]
- **Privacy & isolation** `documented` — Encrypted in transit and at rest; requests consent before sending data to external AI providers. [3]
- **Availability & access** `documented` — Available through messaging and the App Store. [2, 5]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free to start | Free to start | No card required | — | Public usage allowance is not stated. | 1 |
| Pro | $20 / $200 | US App Store purchases | — | The public listing does not identify billing periods or included allowances. | 5 |
| Premium | $200 | US App Store purchase | — | The public listing does not identify the billing period or included allowances. | 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Rene web app at app.rene.co. | 3 |
| iOS | Available | Native iPhone app; iOS 16.4+. | 5 |
| iPadOS | Available | Native iPad app; iPadOS 16.4+. | 5 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Messages on iPhone, WhatsApp, Telegram and the web app. | 2, 3 |
| WhatsApp | Available | Messages on iPhone, WhatsApp, Telegram and the web app. | 2, 3 |
| Telegram | Available | Messages on iPhone, WhatsApp, Telegram and the web app. | 2, 3 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** no public release notes found — Checked: no blog, changelog or updates page on the official site (checked 2026-10-22)

No dated catalogue updates mention this agent yet.

## Sources

1. [Rene product overview](https://rene.co/) — checked 2026-10-21
2. [Rene FAQ](https://rene.co/faq) — checked 2026-10-21
3. [Rene privacy policy](https://rene.co/privacy) — checked 2026-10-21
4. [Rene AI provider list](https://rene.co/ai-providers) — checked 2026-10-21
5. [Rene on the App Store](https://apps.apple.com/us/app/rene-ai-with-friends/id6768238195) — checked 2026-10-21
6. [Rene terms of service](https://rene.co/terms) — checked 2026-10-21

---

# Simular

> A computer-use agent that operates real desktops by seeing the screen, clicking and typing, on your own machine or on a cloud computer you keep.

- **Maker:** Simular Inc.
- **Catalogue type:** Local-first app
- **Website:** https://www.simular.ai
- **Source code:** https://github.com/simular-ai
- **Also known as:** Sai, Agent S
- **Evidence:** 8 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/simular/
- **This record:** https://agentlist.wiki/agents/simular.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Waitlist | documented | 4 |
| Price & usage costs | US$50–500/mo | documented | 4 |
| Included in another plan | Standalone Simular subscription | documented | 4 |
| Where it runs | Your machine or its cloud | documented | 3 |
| Open source & license | Agent S Apache-2.0 | limited | 8 |
| Controls your computer | Yes · Mac, Windows | documented | 3 |
| Its own computer | Cloud · Watch and take over | documented | 2, 3 |
| Works while your computer is off | Keeps working | limited | 2, 4 |
| If the vendor stops | Stops | limited | 4, 8 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 6 |
| Company type | Startup | documented | 5 |
| Launch date | Sep 16, 2026 | documented | 7 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [5]
- **Company stage** `documented` — Series A [5]
- **Disclosed funding** `documented` — US$21.5M · Series A · Dec 2, 2025 [5]
- **Launch date** `documented` — Sep 16, 2026 · Public launch · Sai generally available [7]

**Launch history**

- 2026-09-16 · Public launch · Sai generally available [7]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Operates a real desktop the way a person does, on your own machine or on a cloud computer that keeps your logins, setup and context between sessions [2, 3]
- **Computer use** `documented` — Full desktop control of either the user's own Mac or Windows machine or a Simular-hosted cloud desktop, driving native applications without APIs [2, 3]
- **Controls your computer** `documented` — Drives the user's own Mac or Windows desktop, operating native applications directly. [3]
- **Its own computer** `documented` — An always-on hosted Windows or Mac desktop the user can also watch and take over. [2, 3]
- **Works while your computer is off** `limited` (partial or conditional only) — The hosted always-on computer keeps running independently of the user's own machine, on the tiers that include one. [2, 4]
- **Browser control** `documented` — Drives the browser as one of the desktop applications it operates [3]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Saved workflows rerun nightly, weekly or on a trigger [3]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native macOS app for Apple Silicon and a Windows desktop app [3]
- **iPhone & Android apps** `limited` (partial or conditional only) — No official mobile app; an Android companion exists in the project's repositories [8]
- **Web & messaging** `documented` — Desktop apps, with notifications and approvals through messaging [2, 3]
- **Messaging services** `documented` — iMessage, SMS, Telegram [2]
- **Native hardware** `documented` — A Mac mini can be used as a dedicated machine for the agent, but no hardware is required; the cloud desktop needs none. [3]
- **Developer API** `documented` — A public REST API, with full access gated to the top tier, plus an MCP server [4]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — Built on the Agent S framework and Simulang; the underlying model providers are not named [2, 8]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `documented` — Standalone Simular subscription [4]
- **Price & usage costs** `documented` — Starter US$50/month with a cloud Windows computer, Premium US$200/month with an always-on Windows or Mac computer, Pro US$500/month with full API access, and Enterprise on request. No annual pricing is offered [4]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — The Sai product is a subscription service, though the underlying Agent S framework is Apache-2.0 and would outlive it. The company has not published what happens if the service stops. [4, 8]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Access, correction, deletion and portability are offered by email request, with no in-product export described. [6]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `limited` (partial or conditional only) — The Agent S framework is Apache-2.0; the Sai product itself is not open source [8]
- **Where it runs** `documented` — Runs locally on your own Mac or Windows machine, or on an isolated cloud desktop that Simular hosts [3]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — The agent's virtual machine is a permanent environment where setup, logins, preferences and context persist between sessions [2]
- **Edit & export your data** `limited` (partial or conditional only) — Access, correction, deletion and portability are offered by email request; no in-product export is described [6]
- **Privacy & isolation** `documented` — Cloud work runs on an isolated desktop separate from the user's own machine [3]
- **Availability & access** `documented` — Invite only, with a published route to move up the queue [4]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Starter | US$50/month | Monthly; no annual pricing offered | A cloud Windows computer; Use of your own Windows or Mac machine | Community support; Credit limits apply | 4 |
| Premium | US$200/month | Monthly | An always-on Windows or Mac computer; Higher credits and priority support | Credit limits apply | 4 |
| Pro | US$500/month | Monthly | Unlimited within fair use; Team workflow sharing; Full API access | Fair-use limits apply | 4 |
| Enterprise | Custom | Priced on request | Managed scaling; Single sign-on, role-based access control and SOC 2; Service level agreement | Requires contacting the company | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Native macOS app for Apple Silicon. | 3 |
| Windows | Available | Native Windows desktop app. | 3 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Notifications and approvals arrive by iMessage. | 2 |
| SMS | Available | Notifications and approvals arrive by SMS. | 2 |
| Telegram | Available | Notifications and approvals arrive by Telegram. | 2 |

## Hardware

`optional` Optional hardware — **Dedicated Mac mini**: A Mac mini can be used as a dedicated machine for the agent, but no hardware is required; the cloud desktop needs none. [3]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.simular.ai/blog — Research and product posts, dated (checked 2026-10-22)

- 2026-10-21 · Research · [Agent S4 and evolving agent interfaces](https://www.simular.ai/articles/agent-s4-evolving-the-interfaces-computer-use-agents-work-throughagen) · Simular — A research release reporting that evolving the interfaces a computer-use agent works through lifts a 27B open model to 21.3% on a desktop benchmark, several times the same model with fixed interfaces.

## Sources

1. [Simular official website](https://www.simular.ai) — checked 2026-10-22
2. [Sai product site](https://www.sai.work/) — checked 2026-10-22
3. [Simular Desktop](https://www.simular.ai/simular-desktop) — checked 2026-10-22
4. [Simular pricing](https://www.simular.ai/pricing) — checked 2026-10-22
5. [About Simular](https://www.simular.ai/about) — checked 2026-10-22
6. [Simular privacy policy](https://www.simular.ai/product-sai/privacy) — checked 2026-10-22
7. [Sai becomes generally available](https://www.simular.ai/articles/sai-your-first-robosecretary) — checked 2026-10-22
8. [Agent S repository](https://github.com/simular-ai/Agent-S) — checked 2026-10-22

---

# Skye

> A proactive home-screen assistant that combines calendar, email, finances and health into a ranked feed of contextual cards and briefings.

- **Maker:** Signull Labs
- **Catalogue type:** Hosted agent
- **Website:** https://skyeapp.ai/
- **Source code:** unknown — no public repository recorded
- **Also known as:** Skye Personal AI Agent, Signull Skye
- **Evidence:** 5 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/skye/
- **This record:** https://agentlist.wiki/agents/skye.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · some regions | documented | 2 |
| Price & usage costs | Free | documented | 2 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Provider cloud | documented | 3 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | documented | 4 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 3 |
| Company type | Startup | documented | 5 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [5]
- **Company stage** `documented` — Pre-seed [5]
- **Disclosed funding** `documented` — US$3.58M+ · Pre-seed · Sep 2025 [5]
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Home- and lock-screen widgets surface ranked context from email, calendar, spending, location and Apple Health. [2]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Audio briefings use ElevenLabs voice generation. [3]
- **Background work & triggers** `documented` — Morning briefings and context-sensitive cards appear without opening a chat. [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `documented` — Native iPhone and iPad apps require iOS / iPadOS 26+. [2]
- **Web & messaging** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — OpenAI is the primary provider, with Anthropic, Gemini fallback and Groq for latency-sensitive features. [3]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `documented` — Listed as free on the US App Store. [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — The terms contemplate cessation of the app and promise advance notice where practicable, but set out no data-rescue provision. [4]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A portable copy by emailing the company. Most connected-service data is kept no longer than 90 days, memories until deletion is requested, and deletion requests are handled within 48 hours. [3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Cloud-backed app hosted on Google Cloud, with external model processing. [3]
- **Passwords & credentials** `limited` (partial or conditional only) — Uses an encrypted app-specific password for iCloud Mail; no general credential vault is described. [3]
- **Approval before actions** `documented` — Connections are optional and revocable. Financial integrations are read-only and cannot transfer money. [3]
- **What it remembers** `documented` — Derives memories and relationships from connected sources, including preferences and routines. [3]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — Publisher says AI providers use zero retention and no model training; Skye keeps its own data under a separate retention policy. [3]
- **Availability & access** `documented` — Available on the US App Store. [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | $0 | Free App Store download | Context feed, widgets, briefings and chat over connected services | Public usage quotas are not listed. | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| iOS | Available | Native iPhone app with widgets; requires iOS 26+. | 2 |
| iPadOS | Available | iPad app; requires iPadOS 26+. | 2 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

| Date | Thread | Platform | Kind | Sentiment | Summary |
| --- | --- | --- | --- | --- | --- |
| 2026-10-09 | [Skye: Your life at a glance that lives on your home screen](https://www.reddit.com/r/iosapps/comments/1x1wvcl/skye_your_life_at_a_glance_that_lives_on_your/) | Reddit | announcement | mixed | Launch discussion mixes appreciation with concerns about 90-day data retention and wordy presentation. Several commenters report regional availability gaps; the team explains its retention trade-off. |

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-01 · Release · [Skye rebuilds around a contextual home feed](https://apps.apple.com/us/app/skye-personal-ai-agent/id6748042747) — Version 1.0.8 introduces a rebuilt home-feed experience for the personal assistant.

## Sources

1. [Skye product overview](https://skyeapp.ai/) — checked 2026-10-21
2. [Skye on the App Store](https://apps.apple.com/us/app/skye-personal-ai-agent/id6748042747) — checked 2026-10-21
3. [Skye privacy policy](https://skyeapp.ai/privacy) — checked 2026-10-21
4. [Skye terms](https://skyeapp.ai/terms) — checked 2026-10-21
5. [Skye’s pre-seed financing](https://techcrunch.com/2026/04/27/investors-back-skye-signull-labs-ai-home-screen-app-for-iphone-ahead-of-launch/) — checked 2026-10-21

---

# Thine

> A conversation-memory assistant that records speech and meetings, then uses the shared context for questions, notes and connected-app actions.

- **Maker:** Foyer Tech
- **Catalogue type:** Hosted agent
- **Website:** https://www.thine.com/
- **Source code:** unknown — no public repository recorded
- **Also known as:** Thine AI
- **Evidence:** 4 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/thine/
- **This record:** https://agentlist.wiki/agents/thine.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | Free; $14.99/mo | documented | 1 |
| Included in another plan | iPhone + Mac | documented | 2 |
| Where it runs | Provider cloud | documented | 4 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 4 |
| Remote shut-off | unknown | unknown | — |
| What you keep | None described | documented | 4 |
| Company type | Startup | documented | 4 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [4]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — A shared memory across recordings and chats, with speaker identification and calendar-aware conversations. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Voice & ambient listening** `documented` — Background recording on iPhone, natural voice conversation, and automatic Mac meeting capture using microphone and system audio. [1, 2]
- **Background work & triggers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Native Mac app for Apple Silicon with macOS 15+; Windows is not yet available. [2]
- **iPhone & Android apps** `documented` — Native iPhone app requires iOS 26+; Apple Watch companion requires watchOS 11+. [3]
- **Web & messaging** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — OpenAI, Anthropic and Google provide transcription, summaries, chat and context processing. [4]
- **Bring your own model** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Included in another plan** `documented` — One Thine account and plan carries across iPhone and Mac. [2]
- **Price & usage costs** `documented` — Unlimited recording and transcription are free. Optional Intelligence costs $14.99/month for AI chat, speaker identification and integrations. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [4]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — The privacy policy describes no export right; the terms mention downloading or exporting original recordings only conditionally, "if available". [4]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Stores data with Google Cloud and Cloudflare and uses external AI processing. [4]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Mac capture has a countdown and can be skipped or disabled. External AI processing and memory require consent. [2, 4]
- **What it remembers** `documented` — Recordings and conversations share persistent memory; facts and relationship storage require explicit permission. [1, 4]
- **Edit & export your data** `documented` — Delete recordings and transcripts; request account and memory deletion through support. [4]
- **Privacy & isolation** `documented` — Data encrypted in transit and at rest; the provider says personal data is not used for model training. [4]
- **Availability & access** `documented` — iPhone and Mac apps are available. [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Recorder | Free | Free for life | Unlimited recording and transcription; Retain recordings, transcripts and notes | AI chat and speaker tagging require Intelligence. | 1 |
| Intelligence | $14.99/month | Optional monthly subscription | AI chat about recordings; Speaker identification; 50+ app integrations, including Google, Teams and Outlook | Cancelling removes paid features; recording and transcripts remain available. | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Native app for Apple Silicon Macs with macOS 15+. | 2 |
| iOS | Available | Native iPhone app; iOS 26+. | 3 |
| Apple Watch | Available | Apple Watch companion; watchOS 11+. | 3 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Thine Recorder and Intelligence pricing](https://www.thine.com/recorder) — checked 2026-10-21
2. [Thine for Mac](https://www.thine.com/desktop) — checked 2026-10-21
3. [Thine AI on the App Store](https://apps.apple.com/us/app/thine-ai/id6757556695) — checked 2026-10-21
4. [Thine privacy policy](https://www.thine.com/privacy-policy) — checked 2026-10-21

---

# Tomo

> A personal AI coach in iMessage and iOS that remembers goals, checks in, and builds tools to keep you on track.

- **Maker:** Mapo Labs, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://www.tomo.ai/
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 6 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/tomo/
- **This record:** https://agentlist.wiki/agents/tomo.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1, 3 |
| Price & usage costs | From $19.99/mo | documented | 2 |
| Included in another plan | Standalone Tomo subscription. | documented | 2 |
| Where it runs | Provider cloud | documented | 4 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | unknown | unknown | — |
| Remote shut-off | unknown | unknown | — |
| What you keep | None described | documented | 4 |
| Company type | Startup | documented | 6 |
| Launch date | Jun 25, 2026 | documented | 6 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [6]
- **Company stage** `documented` — Seed [6]
- **Disclosed funding** `documented` — US$5M · Seed · Jun 25, 2026 [6]
- **Launch date** `documented` — Jun 25, 2026 · Public launch [6]

**Launch history**

- 2026-06-25 · Public launch [6]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Goal coaching with proactive check-ins and custom trackers for habits, workouts and budgets. [3]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Can search the web; interactive browser control is not specified. [2]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Sends reminders and initiates check-ins around your goals. [2]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **iPhone & Android apps** `documented` — Native iPhone and iPad app, requiring iOS/iPadOS 18 or later. [3]
- **Web & messaging** `documented` — iMessage and group chats, with a web account dashboard. [1, 2]
- **Messaging services** `documented` — iMessage [1, 2]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Bring your own model** `limited` (partial or conditional only) — Hosted AI responses; bring-your-own model access is not specified. [4]
- **Included in another plan** `documented` — Standalone Tomo subscription. [2]
- **Price & usage costs** `documented` — Base plan starts at $19.99/month; larger messaging allowances and individual promotions are offered. [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — Rights to request access, correction or deletion are described, but no export or portability right is. Deletion requests are processed within 30 days. [4]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Managed service; AI providers process conversations on its behalf. [4]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — Conversation context and goals persist across chats. [3]
- **Edit & export your data** `documented` — Dashboard controls clear messages, reminders and memories; full account deletion can be requested. [2]
- **Privacy & isolation** `documented` — Does not sell conversations; authorized team members can review chats for support, safety and improvement. [4]
- **Availability & access** `documented` — Available through iMessage and the iPhone/iPad app. [1, 3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Base | From $19.99/month | Monthly; cancel in dashboard | Personal AI, reminders and connected calendar/email | Messaging quota is not published; Offers and pricing may vary by account | 2 |
| Higher messaging plans | Shown in account | Account-specific | Larger messaging allowance | Names, quotas and prices are not publicly itemized | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| iOS | Available | Requires iOS 18 or later. | 3 |
| iPadOS | Available | Requires iPadOS 18 or later. | 3 |
| Web | Limited availability | Account, billing and data controls in the web dashboard. | 2 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | iMessage and group chats, with a web account dashboard. | 1, 2 |

## Hardware

`unknown` Not established: No dedicated hardware is specified in the reviewed product pages.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

| Date | Thread | Platform | Kind | Sentiment | Summary |
| --- | --- | --- | --- | --- | --- |
| 2026-03-19 | [Tomo AI subscription?](https://www.reddit.com/r/AI_Agents/comments/1ry23id/tomo_ai_subscription/) | Reddit | community | mixed | Users praise accountability and proactive reminders, while others question pricing, value and access to personal accounts. |

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-06-25 · Launch · [Tomo launches its personal AI in iMessage](https://www.tomo.ai/blog/introducing-tomo) — Mapo Labs introduces a goal-focused personal AI with reminders, group chats and connected calendars.

## Sources

1. [Tomo personal AI](https://www.tomo.ai/) — checked 2026-10-21
2. [Tomo features and pricing FAQ](https://www.tomo.ai/about) — checked 2026-10-21
3. [Tomo iPhone and iPad app](https://apps.apple.com/us/app/tomo-your-personal-ai/id6757726935) — checked 2026-10-21
4. [Tomo privacy policy](https://www.tomo.ai/privacy) — checked 2026-10-21
5. [Tomo terms of service](https://www.tomo.ai/terms) — checked 2026-10-21
6. [Introducing Tomo](https://www.tomo.ai/blog/introducing-tomo) — checked 2026-10-21

---

# Town

> A personal work assistant with shared memory across channels, connected accounts, and recurring background routines.

- **Maker:** Town.com, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://www.town.com
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 13 sources, checked 2026-10-21 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/town/
- **This record:** https://agentlist.wiki/agents/town.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; $15–$49/mo | documented | 4 |
| Included in another plan | Standalone Town individual or team plan | documented | 4 |
| Where it runs | Provider cloud | documented | 1, 5 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | limited | 1, 5 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 12, 13 |
| Company type | Startup | documented | 11 |
| Launch date | Jun 2026 | documented | 11 |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [11]
- **Company stage** `documented` — Series B [11]
- **Disclosed funding** `documented` — US$108M · Series B · Oct 19, 2026 [11]
- **Launch date** `documented` — Jun 2026 · Public launch [11]

**Launch history**

- 2026-06 · Public launch [11]

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Recurring work across connected apps, with per-routine permissions and approvals from your phone [7, 8, 10]
- **Computer use** `limited` (partial or conditional only) — Acts in connected work apps; general desktop control is not specified [1]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `limited` (partial or conditional only) — Web search; interactive browser control is not specified [1]
- **Voice & ambient listening** `documented` — Voice notes, meeting/conversation recording, transcripts and action items in the iOS app [8]
- **Background work & triggers** `documented` — Scheduled and event-triggered routines [3]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Native macOS app; Windows and Linux apps are not specified [9]
- **iPhone & Android apps** `limited` (partial or conditional only) — Native iOS app, available on the App Store; Android app not specified [8]
- **Web & messaging** `documented` — Web, macOS and iOS, plus email, Slack, WhatsApp and Telegram [4, 8, 9]
- **Messaging services** `documented` — WhatsApp, Telegram, Slack, Email [4, 8, 9]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `limited` (partial or conditional only) — Custom MCP servers for agent tools; public agent-control API not specified [10]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Anthropic is primary; OpenAI and Gemini serve selected features and routes [5]
- **Bring your own model** `limited` (partial or conditional only) — Town chooses hosted model providers; personal inference keys are not specified [5]
- **Included in another plan** `documented` — Standalone Town individual or team plan [4]
- **Price & usage costs** `documented` — Free plan includes 30 chats/month; Starter $15/month; Pro from $49/month billed annually [4]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the service depends on the vendor's servers. The company has not published what happens if it shuts down. [1, 5]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A privacy right to a machine-readable copy on request, not a self-serve export. Deleted data is soft-deleted immediately and permanently removed within 30 days. [12, 13]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Provider-hosted service [1, 5]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Ask-before-write by default; safe-actions and allow-all modes are configurable [7]
- **What it remembers** `documented` — Editable profile, preferences, and people notes shared across channels [2]
- **Edit & export your data** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Privacy & isolation** `documented` — Configurable approvals; AI training disabled by default [1, 3]
- **Availability & access** `documented` — Public signup [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free | US$0 | Free | All integrations; weekly briefing; Seven-day Pro trial | 30 chats/month | 4 |
| Starter | US$15/month | Monthly | 1,250 monthly credits; Daily briefing and inbox triage; Two daily routines | 10 people lookups and 15 meeting briefings/month; No credit overage | 4 |
| Pro | US$49/user/month | Annual commitment; US$588/year | 5,000 monthly credits; Unlimited routines and automatic meeting briefings | 50 people lookups/month; Overage: US$0.044/credit | 4 |
| Power | US$99/user/month | Annual commitment; US$1,188/year | 12,000 monthly credits; Priority model routing; higher routine concurrency | 150 people lookups/month; Overage: US$0.038/credit | 4 |
| Power Plus | US$199/user/month | Annual commitment; US$2,388/year | 40,000 monthly credits; Unlimited people lookups; early features | Overage: US$0.032/credit; Email support within two business days | 4 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Web, macOS and iOS, plus email, Slack, WhatsApp and Telegram | 4, 8, 9 |
| macOS | Available | Native macOS app; Windows and Linux apps are not specified | 9 |
| iOS | Available | Native iOS app, available on the App Store; Android app not specified | 8 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| WhatsApp | Available | Web, macOS and iOS, plus email, Slack, WhatsApp and Telegram | 4, 8, 9 |
| Telegram | Available | Web, macOS and iOS, plus email, Slack, WhatsApp and Telegram | 4, 8, 9 |
| Slack | Available | Web, macOS and iOS, plus email, Slack, WhatsApp and Telegram | 4, 8, 9 |
| Email | Available | Web, macOS and iOS, plus email, Slack, WhatsApp and Telegram | 4, 8, 9 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.town.com/updates — Dated release notes with per-entry anchors (checked 2026-10-22)

- 2026-10-21 · Release · [Build routines together in threads](https://www.town.com/updates#release-2026-10-21) · Town — Teammates can be invited into a thread to edit a routine's name, description, instructions or triggers directly.
- 2026-09-30 · Release · [Town 2.0](https://www.town.com/updates#release-2026-09-30) · Town — Turns tasks into shareable threads and changes how the agent requests approvals.
- 2026-09-24 · Release · [CC your Townie](https://www.town.com/updates#release-2026-09-24) · Town — Copying the agent on an email delegates file retrieval and meeting coordination to it.

## Sources

1. [Town getting started](https://www.town.com/docs/getting-started) — checked 2026-10-21
2. [Town context and memory](https://www.town.com/docs/features/context-and-memory) — checked 2026-10-21
3. [Town routines](https://www.town.com/docs/features/routines) — checked 2026-10-21
4. [Town individual plans and credit overages](https://www.town.com/pricing) — checked 2026-10-21
5. [Town AI model providers](https://www.town.com/ai-model-providers) — checked 2026-10-21
6. [Town product updates](https://www.town.com/updates) — checked 2026-10-21
7. [Town security and action approvals](https://www.town.com/features/security) — checked 2026-10-21
8. [Town native iOS app](https://www.town.com/docs/using-town/ios) — checked 2026-10-21
9. [Town integrations and native desktop app](https://www.town.com/integrations) — checked 2026-10-21
10. [Town custom integrations through MCP](https://www.town.com/docs/integrations/mcp-servers) — checked 2026-10-21
11. [Town announces Series B financing and launch history](https://www.town.com/press/series-b) — checked 2026-10-21
12. [Town privacy policy](https://town.com/privacy-policy) — checked 2026-10-22
13. [Town trust and data retention](https://town.com/trust) — checked 2026-10-22

---

# Underdog

> An on-device assistant for email, calendars, notes, dictation, and errands, with locally stored memory.

- **Maker:** Conway Research
- **Catalogue type:** Local-first app
- **Website:** https://underdog.ai
- **Source code:** unknown — no public repository recorded
- **Also known as:** unknown — no other names recorded
- **Evidence:** 6 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/underdog/
- **This record:** https://agentlist.wiki/agents/underdog.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available · invite | documented | 1 |
| Price & usage costs | Free | documented | 3 |
| Included in another plan | Free standalone app | documented | 3 |
| Where it runs | On your device | documented | 1, 2 |
| Open source & license | App license unverified; some model weights Apache-2.0 | limited | 5 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | Stops | limited | 1, 2 |
| If the vendor stops | Partly | limited | 1, 2 |
| Remote shut-off | unknown | unknown | — |
| What you keep | Partial | limited | 1 |
| Company type | Startup | documented | 6 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [6]
- **Company stage** `documented` — First funding round [6]
- **Disclosed funding** `documented` — Undisclosed · First round · Oct 2, 2026 · Led by a16z [6]
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Local model inference, mail, memory and audio processing; optional Naval mode restricts network contact [1, 2]
- **Computer use** `limited` (partial or conditional only) — On-device AI and a browser agent; general desktop control is not specified [1, 2]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `limited` (partial or conditional only) — AI inference runs on the user's own device, so nothing runs while that device is off. [1, 2]
- **Browser control** `documented` — Local browser agent handles bookings and purchases, with approval before consequential actions [2]
- **Voice & ambient listening** `documented` — Local dictation and meeting notes; audio is not uploaded [2]
- **Background work & triggers** `limited` (partial or conditional only) — Automatic email categorization; recurring job scheduler unverified [4]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — macOS on Apple silicon; Windows beta; Linux announced [1]
- **iPhone & Android apps** `documented` — iPhone and iPad on iOS 18+; Android announced [1]
- **Web & messaging** `documented` — Mac, iPhone/iPad, Windows beta [1]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Woof 4B, Underdog 27B, Saluki 27B [1]
- **Bring your own model** `limited` (partial or conditional only) — Ships its own local models; importing other models or inference keys is not specified [1]
- **Included in another plan** `documented` — Free standalone app [3]
- **Price & usage costs** `documented` — Free; revenue comes from purchases it facilitates [3]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Inference runs on the user's own device, so the local app may keep working, though the app licence and update path are unverified. [1, 2]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `limited` (partial or conditional only) — Memory is held locally, but a portable agent-state export is unverified. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `limited` (partial or conditional only) — App license unverified; some model weights Apache-2.0 [5]
- **Where it runs** `documented` — AI inference runs on the user's device; mail connects directly to the user's provider [1, 2]
- **Passwords & credentials** `documented` — Built-in encrypted saved logins on Mac; agent fills passwords without reading them [2]
- **Approval before actions** `documented` — Browser agent asks before buying, paying, sending or deleting [2]
- **What it remembers** `documented` — Local memories can be viewed, edited, and deleted [1]
- **Edit & export your data** `limited` (partial or conditional only) — Local memory; portable agent-state export unverified [1]
- **Privacy & isolation** `documented` — Mac mail, calendar and saved logins are encrypted; key stays in Apple Keychain on the device [2]
- **Availability & access** `documented` — Mac invite access; iOS available without invite; Windows beta [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Underdog | US$0 | Free | Local processing on your computer; No seat or inference subscription | Your computer supplies the compute; Underdog earns a share from purchases it facilitates | 3 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | macOS on Apple silicon; Windows beta; Linux announced | 1 |
| Windows | Beta | macOS on Apple silicon; Windows beta; Linux announced | 1 |
| Linux | Coming soon | macOS on Apple silicon; Windows beta; Linux announced | 1 |
| iOS | Available | iPhone and iPad on iOS 18+; Android announced | 1 |
| iPadOS | Available | iPhone and iPad on iOS 18+; Android announced | 1 |
| Android | Coming soon | iPhone and iPad on iOS 18+; Android announced | 1 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-15 · Release · [Underdog adds a Windows beta](https://underdog.ai/ship) — The official release log announces Windows availability for its local personal assistant.
- 2026-10-13 · Release · [Underdog adds purchases through Link](https://underdog.ai/ship) — Version 1.4.1 introduces purchases using a one-time Link card approved by the user.

## Sources

1. [Underdog official product site](https://underdog.ai/) — checked 2026-10-21
2. [Underdog security](https://underdog.ai/security) — checked 2026-10-21
3. [Underdog pricing](https://underdog.ai/pricing) — checked 2026-10-21
4. [Underdog release log](https://underdog.ai/ship) — checked 2026-10-21
5. [Conway Research Underdog-27B-1.0 model repository](https://huggingface.co/ConwayResearch/Underdog-27B-1.0) — checked 2026-10-21
6. [Andreessen Horowitz leads Conway Research’s first funding round](https://a16z.com/announcement/investing-in-conway-the-creator-of-underdog/) — checked 2026-10-21

---

# Vora Aegis

> A local AI appliance you own: an assistant called a Guardian builds an encrypted database of your digital life on the device and runs open models on it. Announced for Spring 2027; reservations only.

- **Maker:** Vora
- **Catalogue type:** Local-first app
- **Website:** https://vora.io
- **Source code:** unknown — no public repository recorded
- **Also known as:** Vora, Aegis, Guardian
- **Evidence:** 5 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/vora/
- **This record:** https://agentlist.wiki/agents/vora.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Reservation · some regions | documented | 1, 3, 5 |
| Price & usage costs | US$2,499 device | documented | 2, 3, 5 |
| Included in another plan | One-time purchase; no subscription required | documented | 2, 3 |
| Where it runs | Your own appliance | documented | 2, 3 |
| Open source & license | Open source stated; no repository linked | limited | 2, 3 |
| Controls your computer | No | documented | 1, 3 |
| Its own computer | Own hardware · no remote view | documented | 2, 3 |
| Works while your computer is off | Keeps working | limited | 2, 3 |
| If the vendor stops | Keeps working | documented | 2, 3 |
| Remote shut-off | Not possible | documented | 2, 3 |
| What you keep | Self-serve | documented | 2, 3 |
| Company type | Startup | documented | 4 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [4]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — A desktop appliance you own that runs open models on-device, split so the layer holding your data has no unrestricted internet access and the internet-facing layer holds no private data [1, 2]
- **Computer use** `limited` (partial or conditional only) — Acts from its own appliance through connected accounts and a secure browser; it does not control your Mac or PC, and no remote interface onto the device's own screen is described [1, 3]
- **Controls your computer** `documented` — Nothing runs on your Mac or PC. The Aegis appliance is a separate machine that does the work itself. [1, 3]
- **Its own computer** `documented` — The Aegis appliance is the agent's computer and it has no screen or keyboard; you reach it through the phone and web apps rather than viewing its desktop. [2, 3]
- **Works while your computer is off** `limited` (partial or conditional only) — The appliance does the work and stays on by itself; your phone pairs with it and is not what runs the models. [2, 3]
- **Browser control** `documented` — A secure browser runs errands on the internet from the internet-facing layer, which holds no private data [1, 2]
- **Voice & ambient listening** `documented` — No ambient listening by design; the company contrasts Aegis with always-on speakers and says you talk to it through the app when you choose. A voice interface is not described [3]
- **Background work & triggers** `limited` (partial or conditional only) — Runs errands sent out to the internet and keeps a current brief in the app; scheduled or event-triggered routines are not specified [3]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — No native desktop app is described; deeper work such as connecting accounts happens in the web app from a laptop [3]
- **iPhone & Android apps** `limited` (partial or conditional only) — A phone app pairs the device and handles quick questions and notifications; iPhone and Android availability is not separately specified, and full phone-desktop parity is described as growing after launch [3]
- **Web & messaging** `documented` — A phone app and a web app. The device itself has no screen or keyboard [3]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `limited` (partial or conditional only) — A compact desktop appliance, 225mm x 185mm x 50mm and 2.2 lbs, with an AMD Ryzen AI Max+ 395, Radeon 8060S, 64GB unified memory, a 1TB SSD expandable through two M.2 sockets, and the Harpē security module. The device is required: it is the product. Announced to ship Spring 2027, and the site states its images are renders of the production enclosure [1, 2, 3]
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Ships running Gemma 4 26B entirely on the device; the specifications cite over 100 tokens per second for open models of that size, naming Qwen 3.6 35B and Ornith 1.5 35B [2, 3]
- **Bring your own model** `documented` — The model is the owner's to choose and can be replaced with newer open models as they are released [3]
- **Included in another plan** `documented` — Bought outright; the company states the machine runs the full experience with no subscription, and all three optional plans are future development that will not be available at launch [2, 3]
- **Price & usage costs** `documented` — US$2,499 Super Early Bird reservation with a US$50 refundable deposit; a published ladder runs to US$2,799 Early Bird and US$2,999 retail. Optional post-launch plans are US$20/month Guardian Care, US$20/month Aegis Pro and US$100/month Aegis Pro+ [2, 3, 5]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — The company states the appliance keeps working without its servers, needs no subscription, and cannot be switched off by an activation server. [2, 3]
- **Remote shut-off** `documented` — No activation server can disable it; the owner holds the keys and the machine runs without Vora. [2, 3]
- **What you keep** `documented` — Everything exports in open formats at any time, with no lock-in or exit fee described. [2, 3]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `limited` (partial or conditional only) — The specifications state open source top to bottom, and the company says updates are checked against published source; no public repository is linked from the site, so the code could not be inspected [2, 3]
- **Where it runs** `documented` — Runs on the owner's appliance on EnclaveOS, a hardened security-first operating system. Account data flows from the provider straight to the device rather than through a Vora server. Optional frontier-model credits would send only approved content off-device, and are not available at launch [2, 3]
- **Passwords & credentials** `limited` (partial or conditional only) — Sign-in keys for connected accounts are held on the device, and the Harpē security module holds the encryption keys plus a hardware wallet for payments; a general-purpose password manager is not described [2, 3]
- **Approval before actions** `documented` — Anything consequential is drafted and waits for approval before it is sent; NFC tap-to-verify provides multi-signature confirmation of sensitive data and of payments above a threshold the owner sets [2, 3]
- **What it remembers** `documented` — One encrypted database on the device holds mail, messages, photos, calendar, files, financial and health accounts; conversation history imports from ChatGPT, Claude and Gemini. One owner per device, with household modes not yet offered [1, 3]
- **Edit & export your data** `documented` — Everything exports in open formats at any time, and connected data can be deleted from Settings; the company states there is no lock-in and the device keeps working without its servers [2, 3]
- **Privacy & isolation** `documented` — Harpē, a hardware security module with dual secure elements, holds keys locally; the split-agent design is presented as structural protection against prompt-injection exfiltration. No training on customer data, signed updates verified against published source with automatic rollback, and a recovery phrase the company states it cannot recover [2, 3]
- **Availability & access** `documented` — Not yet available. Reservations only: the US$50 Super Early Bird deposit holds one of 200 units, refundable until the unit ships. Aegis is stated to ship Spring 2027, launching in the United States, Canada and the European Union [1, 3, 5]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Aegis Super Early Bird | US$2,499 | One-time purchase; US$50 refundable deposit to reserve | Vora Aegis appliance, owned outright; AMD Ryzen AI Max+ 395, Radeon 8060S, 64GB unified memory, 1TB SSD; Harpē hardware security module with dual secure elements; Two-year hardware warranty, not tied to any subscription | One of 200 units at this tier; Founders at US$1,999 is sold out; Ladder continues at US$2,799 Early Bird and US$2,999 retail; Ships Spring 2027; deposit refundable until the unit ships | 2, 3, 5 |
| Guardian Care | US$20/month | Optional subscription; not available at launch | Complete off-site encrypted backup under keys only the owner holds; Restores a replacement device after loss or hardware failure | Future development; the company states it is not available at launch; Recovery phrase is the only key and cannot be recovered by Vora | 2, 3 |
| Aegis Pro | US$20/month | Optional subscription; not available at launch | Monthly allotment of frontier cloud-model credits for heavier work | Future development; the company states it is not available at launch; Only content approved for a task leaves the device | 2, 3 |
| Aegis Pro+ | US$100/month | Optional subscription; not available at launch | Larger allotment of frontier cloud-model credits; Guardian Care included | Future development; the company states it is not available at launch | 2, 3 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Coming soon | A web app is described for connecting accounts and managing settings; it ships with the device in Spring 2027 | 3 |
| Mobile app | Coming soon | A phone app pairs the device and handles quick questions and notifications; iPhone and Android are not separately specified, and it ships with the device in Spring 2027 | 3 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`announced` Announced, not shipping — **Vora Aegis**: A compact desktop appliance, 225mm x 185mm x 50mm and 2.2 lbs, with an AMD Ryzen AI Max+ 395, Radeon 8060S, 64GB unified memory, a 1TB SSD expandable through two M.2 sockets, and the Harpē security module. The device is required: it is the product. Announced to ship Spring 2027, and the site states its images are renders of the production enclosure [1, 2, 3]

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

No dated catalogue updates mention this agent yet.

## Sources

1. [Vora Aegis official website](https://vora.io/) — checked 2026-10-22
2. [Vora Aegis technical specifications](https://vora.io/technical-specifications/) — checked 2026-10-22
3. [Vora Aegis frequently asked questions](https://vora.io/faq/) — checked 2026-10-22
4. [About Vora: team and principles](https://vora.io/about/) — checked 2026-10-22
5. [Vora Aegis Reservation, $2,499 Super Early Bird](https://shop.vora.io/products/vora-aegis-vip-deposit) — checked 2026-10-22

---

# Wabi

> A shared personal agent for friends and family, with group chats and custom apps built inside each conversation.

- **Maker:** Wabi, Inc.
- **Catalogue type:** Hosted agent
- **Website:** https://wabi.ai/
- **Source code:** unknown — no public repository recorded
- **Also known as:** Wabi 2.0
- **Evidence:** 6 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/wabi/
- **This record:** https://agentlist.wiki/agents/wabi.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 3 |
| Price & usage costs | Free download | limited | 5 |
| Included in another plan | unknown | unknown | — |
| Where it runs | Cloud + Mac context | documented | 4 |
| Open source & license | unknown | unknown | — |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Stops | documented | 4 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 4 |
| Company type | Startup | documented | 6 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [6]
- **Company stage** `documented` — Pre-seed [6]
- **Disclosed funding** `documented` — US$20M · Pre-seed · Nov 5, 2025 [6]
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — A shared agent that turns group conversations into plans, trackers, bookings and custom mini-apps. [1, 5]
- **Computer use** `limited` (partial or conditional only) — Hosted browser sessions; Mac app can read permitted screen and file context, but general desktop control is not specified [4]
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Hosted browser sessions complete tasks on websites [4]
- **Voice & ambient listening** `documented` — Voice input and voice/video calls; Mac screen sharing during a call [4]
- **Background work & triggers** `limited` (partial or conditional only) — Proactive updates; optional location-triggered help; recurring schedules unverified [4, 5]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `limited` (partial or conditional only) — Native macOS app; Windows and Linux apps not specified [3, 4]
- **iPhone & Android apps** `documented` — Native iOS and Android apps [3]
- **Web & messaging** `documented` — iOS, Android, Mac; group conversations [1, 3]
- **Messaging services** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `limited` (partial or conditional only) — Google, Anthropic and OpenAI; specific models not listed [4]
- **Bring your own model** `limited` (partial or conditional only) — Uses Google, Anthropic and OpenAI providers; personal inference keys are not specified [4]
- **Included in another plan** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Price & usage costs** `limited` (partial or conditional only) — Free App Store download; no public tier or usage-limit table. [5]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — The terms reserve the right to modify or discontinue all or any portion of the service at any time and without notice; data may transfer to a successor in a merger, bankruptcy or liquidation. [4]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A right to request a copy in a structured, commonly used, machine-readable format. Deleted accounts are deleted or anonymised within 30 days. [4]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Cloud service and hosted browsers, with permitted local context from the Mac app [4]
- **Passwords & credentials** `documented` — Built-in Wallet encrypts saved credentials and releases them after device authentication [4]
- **Approval before actions** `documented` — Checks before important actions; Mac Messages sends require approval; Wallet requires device authentication [4, 5]
- **What it remembers** `documented` — Persistent personal memory; retained while account is active [2, 4]
- **Edit & export your data** `limited` (partial or conditional only) — Can import conversation history and preferences from other assistants; full state export not specified [4]
- **Privacy & isolation** `documented` — Approval boundaries; connected-account content excluded from model training [4, 5]
- **Availability & access** `documented` — iOS, Android, and Mac downloads listed [3]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Wabi | Free download | App Store | Group conversations and generated mini-apps | Public usage limits are not listed | 5 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Native macOS app; Windows and Linux apps not specified | 3, 4 |
| iOS | Available | Native iOS and Android apps | 3 |
| Android | Available | Native iOS and Android apps | 3 |

## Messaging services

unknown — no messaging service is recorded for this agent.

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

| Date | Thread | Platform | Kind | Sentiment | Summary |
| --- | --- | --- | --- | --- | --- |
| 2026-10-20 | [Replika 2? Wabi](https://www.reddit.com/r/replika/comments/1xaid4i/replika_2_wabi/) | Reddit | community | positive | Early users welcome the agentic update and wider platform access, and describe building apps together. |

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-09-28 · Release · [Wabi 3.0 becomes an AI messenger](https://apps.apple.com/us/app/wabi-ai-assistant/id6747768928) — The release adds group conversations, task assistance, proactive updates, and generated interfaces inside chats.

## Sources

1. [Wabi personal assistant](https://wabi.ai/) — checked 2026-10-21
2. [Wabi manifesto](https://wabi.ai/manifesto) — checked 2026-10-21
3. [Download Wabi](https://wabi.ai/download) — checked 2026-10-21
4. [Wabi privacy notice](https://wabi.ai/privacy) — checked 2026-10-21
5. [Wabi official App Store listing](https://apps.apple.com/us/app/wabi-ai-assistant/id6747768928) — checked 2026-10-21
6. [Andreessen Horowitz invests in Wabi](https://a16z.com/announcement/investing-in-wabi/) — checked 2026-10-21

---

# ZeroClaw

> A Rust agent runtime combining persistent operation, messaging, tools, and hardware access.

- **Maker:** ZeroClaw Labs
- **Catalogue type:** Self-hosted runtime
- **Website:** https://www.zeroclaw.com
- **Source code:** https://github.com/zeroclaw-labs/zeroclaw
- **Also known as:** unknown — no other names recorded
- **Evidence:** 8 sources, checked 2026-10-21 to 2026-10-21
- **Profile:** https://agentlist.wiki/agent/zeroclaw/
- **This record:** https://agentlist.wiki/agents/zeroclaw.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 1 |
| Price & usage costs | Free; you pay models | documented | 1 |
| Included in another plan | Codex subscription or Claude Max setup token | documented | 1 |
| Where it runs | Self-hosted | documented | 1 |
| Open source & license | MIT OR Apache-2.0 | documented | 1 |
| Controls your computer | unknown | unknown | — |
| Its own computer | unknown | unknown | — |
| Works while your computer is off | unknown | unknown | — |
| If the vendor stops | Keeps working | documented | 1 |
| Remote shut-off | Not possible | documented | 1 |
| What you keep | Partial | documented | 1 |
| Company type | unknown | unknown | — |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — Event-triggered procedures can react to MQTT, webhooks, timers, and hardware events. [1]
- **Computer use** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Controls your computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Its own computer** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Works while your computer is off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Browser control** `documented` — Opt-in Chrome/Chromium automation; actions normally go through the approval policy. [2]
- **Voice & ambient listening** `documented` — Phone calls, local wake-word listening, and spoken output through configured speech providers. [7]
- **Background work & triggers** `documented` — Cron and event-triggered operating procedures. [1]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Tauri desktop apps with macOS, Windows, and Linux installers. [3, 4]
- **iPhone & Android apps** `limited` (partial or conditional only) — Experimental Android command-line runtime via Termux/ADB; native mobile app not specified. [5]
- **Web & messaging** `documented` — Discord, Telegram, Matrix, email, webhooks, voice, CLI. [1]
- **Messaging services** `documented` — Telegram, Discord, Matrix, Email [1]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — Self-hosted Gateway REST API with bearer authentication and a built-in OpenAPI explorer. [6]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Anthropic, OpenAI, Ollama, and compatible endpoints. [1]
- **Bring your own model** `documented` — Own provider credentials and local models such as Ollama. [1]
- **Included in another plan** `documented` — Supports Codex subscription authentication and Claude Max setup-token authentication. [1]
- **Price & usage costs** `documented` — Open-source software with no software subscription. Supply your own model access and deployment; their costs and limits vary. [1]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `documented` — Open-source software you run yourself, so it keeps running if the project stops; it still depends on whichever model provider you point it at. [1]
- **Remote shut-off** `documented` — Runs on hardware you control, with no vendor activation or licence check described. [1]
- **What you keep** `documented` — State lives on hardware you control, but a documented full export and restore path is not established. [1]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `documented` — MIT OR Apache-2.0 [1]
- **Where it runs** `documented` — Self-hosted; background system service. [1]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `documented` — Supervised mode asks for medium-risk actions and blocks high-risk actions by default. [8]
- **What it remembers** `documented` — SQLite and embeddings. [1]
- **Edit & export your data** `limited` (partial or conditional only) — Single Rust binary and configurable providers. Full state export / restore not verified. [1]
- **Privacy & isolation** `documented` — Supervised defaults, workspace boundaries, approval gates, OS sandboxes. [1]
- **Availability & access** `documented` — Available for self-hosting. [1]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Self-hosted | Free software | Model access and hosting are supplied separately. | MIT OR Apache-2.0 open-source license; Own provider credentials and local models such as Ollama. | Usage limits and running costs depend on your chosen provider and deployment. | 1 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| macOS | Available | Tauri desktop apps with macOS, Windows, and Linux installers. | 3, 4 |
| Windows | Available | Tauri desktop apps with macOS, Windows, and Linux installers. | 3, 4 |
| Linux | Available | Tauri desktop apps with macOS, Windows, and Linux installers. | 3, 4 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| Telegram | Available | Discord, Telegram, Matrix, email, webhooks, voice, CLI. | 1 |
| Discord | Available | Discord, Telegram, Matrix, email, webhooks, voice, CLI. | 1 |
| Matrix | Available | Discord, Telegram, Matrix, email, webhooks, voice, CLI. | 1 |
| Email | Available | Discord, Telegram, Matrix, email, webhooks, voice, CLI. | 1 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** unknown — no update source has been recorded for this agent.

- 2026-10-18 · Release · [ZeroClaw v0.8.6](https://github.com/zeroclaw-labs/zeroclaw/releases/tag/v0.8.6) · ZeroClaw Labs — 573 commits from 74 contributors. Command-line release archives now carry the WebAssembly plugin host, the plugin registry offers channel and tool packages, and the release adds authenticated principals.

## Sources

1. [ZeroClaw README](https://raw.githubusercontent.com/zeroclaw-labs/zeroclaw/master/README.md) — checked 2026-10-21
2. [ZeroClaw browser automation](https://github.com/zeroclaw-labs/zeroclaw/blob/master/docs/book/src/tools/browser.md) — checked 2026-10-21
3. [ZeroClaw desktop app](https://github.com/zeroclaw-labs/zeroclaw/blob/master/apps/tauri/TESTING.md) — checked 2026-10-21
4. [ZeroClaw v0.8.6 release assets](https://github.com/zeroclaw-labs/zeroclaw/releases/tag/v0.8.6) — checked 2026-10-21
5. [ZeroClaw Android runtime](https://github.com/zeroclaw-labs/zeroclaw/blob/master/docs/book/src/hardware/android-setup.md) — checked 2026-10-21
6. [ZeroClaw Gateway HTTP API](https://github.com/zeroclaw-labs/zeroclaw/blob/master/docs/book/src/gateway/api.md) — checked 2026-10-21
7. [ZeroClaw voice and telephony](https://github.com/zeroclaw-labs/zeroclaw/blob/master/docs/book/src/channels/voice.md) — checked 2026-10-21
8. [ZeroClaw security model](https://github.com/zeroclaw-labs/zeroclaw/blob/master/docs/book/src/security/model.md) — checked 2026-10-21

---

# Zo Computer

> A personal always-on cloud computer with a real filesystem, reachable over chat, text or email, which runs automations and can also host sites and services.

- **Maker:** Zo Computer Company
- **Catalogue type:** Hosted agent
- **Website:** https://zo.computer
- **Source code:** unknown — no public repository recorded
- **Also known as:** Zo
- **Evidence:** 12 sources, checked 2026-10-22 to 2026-10-22
- **Profile:** https://agentlist.wiki/agent/zo/
- **This record:** https://agentlist.wiki/agents/zo.md

## Key facts

| Field | Value | Status | Sources |
| --- | --- | --- | --- |
| Availability & access | Available | documented | 2 |
| Price & usage costs | US$18–200/mo after trial | documented | 2 |
| Included in another plan | Standalone Zo subscription | documented | 2 |
| Where it runs | Zo cloud, plus your machine | documented | 3, 5 |
| Open source & license | unknown | unknown | — |
| Controls your computer | Yes · Mac, Windows, Linux | documented | 5 |
| Its own computer | Cloud · Watch and take over | documented | 3, 4 |
| Works while your computer is off | Keeps working | documented | 3 |
| If the vendor stops | Stops | limited | 3 |
| Remote shut-off | unknown | unknown | — |
| What you keep | On request | documented | 2, 11 |
| Company type | Startup | documented | 12 |
| Launch date | unknown | unknown | — |

Status marks — `documented`: Supported by official documentation; not independently tested. `limited`: Only partial or conditional support established. `unknown`: Not established by reviewed sources; does not mean unsupported.

## Company & launch

*Who makes it and when it arrived. Company type and funding are context, not a quality score: an open-source project and a public company carry different risks, not better or worse ones.*

- **Company type** `documented` — Startup [12]
- **Company stage** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Disclosed funding** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Launch date** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".

**Launch history**

unknown — no dated launch event is recorded.

## What it can do

*What this agent can actually operate — a machine of its own, yours, or neither. Computer control is three separate questions, and most products answer only some of them.*

- **What stands out** `documented` — An always-on cloud computer of your own with a 100 GB filesystem that keeps running with your laptop closed, and can host sites and services on custom domains [1, 2, 3]
- **Computer use** `documented` — Runs its own Chromium on your Zo workspace; the desktop app's Devices feature can additionally drive your physical computer, with separate opt-in for browser, shell and file access [4, 5]
- **Controls your computer** `documented` — The desktop app's Devices feature runs commands on your physical machine, with separate opt-in for browser, shell and file access. [5]
- **Its own computer** `documented` — An always-on cloud computer with its own Chromium, streamed into a browser tab where you can watch and take over. [3, 4]
- **Works while your computer is off** `documented` — Automations run on the Zo cloud computer even while your laptop is closed. [3]
- **Browser control** `documented` — Its own Chromium streamed into a browser tab, where you sign in once and it can then act on those sites as you [4]
- **Voice & ambient listening** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Background work & triggers** `documented` — Automations are described as a skill with a clock, triggered by events such as incoming email or file changes, by daily, weekly, monthly or custom schedules, or manually, and run even while your laptop is closed [3]

## Apps & access

*Where you reach it and on what. A messaging integration is not a native app, so the two are tracked apart; an app listed here is a real first-party client.*

- **Native desktop apps** `documented` — Desktop apps for macOS, Windows and Linux with file sync [7]
- **iPhone & Android apps** `limited` (partial or conditional only) — Mobile access is through the web; no native mobile app is described [1]
- **Web & messaging** `documented` — Web interface and desktop apps, plus messaging channels and a personal email address [1, 6]
- **Messaging services** `documented` — iMessage, SMS, Telegram, Slack, Discord, Email [6]
- **Native hardware** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Developer API** `documented` — A public API with a token from advanced settings [10]

## Models & plans

*What you pay and what runs underneath. Where an agent is bundled into a larger subscription, the price shown is that subscription's — there is no separate fee for the agent.*

- **Models & providers** `documented` — Anthropic, OpenAI, Google, xAI, DeepSeek, Moonshot, Z.ai, ByteDance, MiniMax and AssemblyAI; every model is unlocked on paid tiers [2, 8]
- **Bring your own model** `documented` — Connect your own keys from OpenAI-compatible, Anthropic-compatible or Groq endpoints, and sign in to Codex and Gemini subscriptions [9]
- **Included in another plan** `documented` — Standalone Zo subscription [2]
- **Price & usage costs** `documented` — Free fourteen-day trial, then Basic US$18/month with US$10 of AI credits, Pro US$64/month with US$40, and Ultra US$200/month with US$100. Every tier has a 100 GB disk [2]

## Vendor lock-in

*How hard it is to leave, and what survives if the company does. Almost no hosted vendor publishes a shutdown commitment, so an answer here is usually read from the architecture rather than from a promise.*

- **If the vendor stops** `limited` (partial or conditional only) — Provider-hosted, so the computer and its filesystem depend on the vendor. The company has not published what happens if it shuts down. [3]
- **Remote shut-off** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What you keep** `documented` — A portability right exercised by contacting the company, and ZIP archives from support after a trial; individual files are directly downloadable. [2, 11]

## Ownership & control

*What it asks before acting, what it remembers, and who else can see it. These are the questions that decide whether you would hand it your inbox.*

- **Open source & license** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Where it runs** `documented` — Primarily an always-on computer in Zo's cloud; the desktop app can also connect your own machine [3, 5]
- **Passwords & credentials** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **Approval before actions** `unknown` — Not specified — no reviewed source answers this, which is a gap in the catalogue rather than a "no".
- **What it remembers** `documented` — A persistent 100 GB filesystem with chat history, personas and rules; improved memory that remembers your work across sessions is a paid-tier feature, and files are directly browsable and editable [1, 2]
- **Edit & export your data** `limited` (partial or conditional only) — A right to receive your data in a portable format by contacting the company, and ZIP archives from support after a trial ends; files themselves are directly downloadable [2, 11]
- **Privacy & isolation** `documented` — The Devices feature requires separate opt-in for browser, shell and file access on your physical computer [5]
- **Availability & access** `documented` — Open self-serve sign-up with a fourteen-day trial and no credit card required; after the trial, ordinary computer access ends without a subscription [2]

## Plans

| Plan | Price | Billing | Includes | Limits | Sources |
| --- | --- | --- | --- | --- | --- |
| Free trial | US$0 | Fourteen days, no credit card | 100 GB disk; One hosted service | Computer sleeps when you step away; Limited AI and no improved memory; Ordinary computer access ends after fourteen days | 2 |
| Basic | US$18/month | Monthly | Always on, with improved memory; US$10 a month of AI credits; Five services and three custom domains | 2 vCPU and 8 GB memory | 2 |
| Pro | US$64/month | Monthly | US$40 a month of AI credits; Ten services and five custom domains; Priority support | 4 vCPU and 16 GB memory | 2 |
| Ultra | US$200/month | Monthly | US$100 a month of AI credits; Fifty services and ten custom domains | 8 vCPU and 32 GB memory | 2 |

## Native apps and web access

*Messaging services are channels, not native apps, and are listed separately below.*

| Platform | Status | Detail | Sources |
| --- | --- | --- | --- |
| Web | Available | Primary web interface. | 1 |
| macOS | Available | Desktop app with file sync and optional device control. | 5, 7 |
| Windows | Available | Desktop app with file sync and optional device control. | 7 |
| Linux | Available | Desktop app for Linux, x64 and ARM. | 7 |

## Messaging services

| Service | Status | Detail | Sources |
| --- | --- | --- | --- |
| iMessage | Available | Listed on the Zo messaging page. | 6 |
| SMS | Available | Listed among Zo messaging channels. | 6 |
| Telegram | Available | Listed among Zo messaging channels. | 6 |
| Slack | Available | Listed among Zo messaging channels. | 6 |
| Discord | Available | Listed among Zo messaging channels. | 6 |
| Email | Available | Each account gets a personal address at zo.computer. | 6 |

## Hardware

`unknown` — no dedicated hardware is established by the reviewed sources.

## Data reach

*What personal data this agent can reach, and whether anyone else could reach the same source.*

unknown — no first-party or exclusive data reach is recorded for this agent.

## Discussion

unknown — no dated public threads are recorded for this agent.

## Feedback snapshot

unknown — no attributed feedback snapshot is recorded for this agent.

## Updates

**Where updates come from:** https://www.zo.computer/updates — Monthly update posts and a weekly changelog (checked 2026-10-22)

- 2026-09-30 · Release · [What's new in Zo](https://www.zo.computer/updates/2026-09-30) · Zo Computer — Adds several new models, a reasoning-effort setting, in-chat rendering of MCP apps, a concurrency setting for the ask endpoint, and reorganised billing.

## Sources

1. [Zo Computer official website](https://zo.computer) — checked 2026-10-22
2. [Zo pricing](https://zo.computer/pricing) — checked 2026-10-22
3. [Zo automations](https://www.zo.computer/automations) — checked 2026-10-22
4. [Zo browser guide](https://www.zo.computer/guide/browser) — checked 2026-10-22
5. [Zo devices guide](https://www.zo.computer/guide/devices) — checked 2026-10-22
6. [Zo messaging channels](https://www.zo.computer/guide/messaging-channels) — checked 2026-10-22
7. [Zo desktop app](https://www.zo.computer/guide/desktop) — checked 2026-10-22
8. [Zo models](https://www.zo.computer/models) — checked 2026-10-22
9. [Zo bring your own key](https://www.zo.computer/guide/byok) — checked 2026-10-22
10. [Zo API](https://www.zo.computer/guide/api) — checked 2026-10-22
11. [Zo privacy policy](https://www.zo.computer/privacy) — checked 2026-10-22
12. [Zo team](https://www.zo.computer/team) — checked 2026-10-22