# 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