Hermes Agent on GitHub

Where to Find Hermes on GitHub

Hermes Agent is maintained by Nous Research and published under the NousResearch/hermes-agent repository. Everything lives there: the full Python source, documentation, Docker configuration, tagged releases, and the issue tracker.

Key things you will find in the repository:

• Source code — the gateway server, platform integrations (Telegram, Discord, Slack, and an OpenAI-compatible API server on port 8642), tool implementations for web search, image generation, vision, audio transcription, and shell execution
• Configuration reference — config.yaml schema, environment variable names, and platform setup instructions
• Releases tab — versioned tags you can pin in your Docker image or install command
• Issue tracker — bug reports, feature requests, and upstream discussions
• Skills directory — optional skills (including a free DuckDuckGo search skill that activates automatically when no paid search key is set)

If you want to star or fork the project, watch for new releases, or contribute a pull request, that repository is your starting point.

Cloning and Building Hermes from Source

The exact setup commands are documented in the repository README and may change between releases. The general pattern looks like:

# Clone the repository
git clone https://github.com/NousResearch/hermes-agent.git
cd hermes-agent

# Checkout a specific release tag — never run main in production
git checkout <release-tag>

# See the repo README for the exact install and run commands
# (Python version requirements, dependency install, config setup, etc.)

Before running any commands, read the README in the tag you checked out. Install instructions, Python version requirements, and environment variable names can differ between releases.

Hermes also ships an official Docker image, which is the recommended way to run it in production. The Docker path avoids Python environment conflicts and makes version pinning straightforward.

Releases, Tags, and Version Pinning

Hermes uses tagged releases (for example v2026.4.16) rather than a rolling main branch for production use. The Releases page lists every version with its changelog.

When running Hermes yourself, always:

• Pin your Docker image to a specific release tag, not :latest or :main
• Read the release notes before upgrading — tool APIs, config field names, and platform behavior can change between versions
• Test a new tag in a non-production environment before rolling it to live users

Running main in production means you get unreviewed commits that may break config compatibility or introduce regressions. A pinned tag gives you a reproducible, rollback-able baseline.

Issues and Feature Requests

The GitHub issue tracker is the right place to report bugs in Hermes itself — unexpected tool behavior, config parsing failures, upstream API compatibility issues, or feature requests for the core agent.

If your issue is specific to hosting Hermes on OpenClaw Launch — deployment errors, the Connect Platforms UI, billing, or container management — reach out to [email protected] instead. We track those separately and can act on them faster than a GitHub thread.

When filing an upstream issue, include the exact release tag you are running, your relevant config.yaml sections (redact any tokens), and the full error output. Nos Research contributors triage faster with complete reproduction details.

How OpenClaw Launch Uses Hermes

OpenClaw Launch runs a pinned Hermes release tag inside a managed Docker container on dedicated infrastructure. When you deploy a Hermes instance through Launch, you skip the entire clone-build-configure-host sequence: no git checkout, no Python environment setup, no server provisioning, no TLS configuration. Launch handles the container lifecycle, version tracking, and platform wiring automatically. If Nous Research ships a new Hermes release that passes our testing, we update the pinned tag and you get the upgrade on your next deploy without touching any config files.

Frequently Asked Questions

Is Hermes Agent open source?

Yes. The full source is public at github.com/NousResearch/hermes-agent under the license stated in the repository. You can clone it, fork it, and contribute pull requests.

What platforms does Hermes support?

Hermes supports Telegram, Discord, Slack, and an OpenAI-compatible API server (port 8642 by default, enabled via API_SERVER_ENABLED=true in ~/.hermes/.env). The API server supports streaming via SSE, session continuity via previous_response_id or a named conversation, and the standard /v1/chat/completions and /v1/responses endpoints. Additional platforms are configured via config.yaml.

What tools does Hermes include out of the box?

The upstream repo includes web search (Tavily, Exa, Parallel, Firecrawl, or a free DuckDuckGo fallback), image generation via FAL.ai, vision inputs, audio transcription, and shell execution across local, Docker, SSH, and other execution environments. An approval policy (approvals.mode: manual, smart, or off) controls whether tool calls require confirmation.

How do I report a bug in Hermes?

File an issue at github.com/NousResearch/hermes-agent/issues. Include your release tag, the relevant config sections (redact tokens), and the full error output. If the bug only occurs when running on OpenClaw Launch, contact [email protected] first so we can determine whether it is a hosting issue or an upstream bug.

Does OpenClaw Launch stay current with Hermes releases?

We track new Hermes releases, test them against our platform, and update the pinned tag when the new version is stable. We do not run :latest — each version is validated before being made available to instances.

Related Guides

• What Is Hermes Agent? — Overview of Hermes capabilities and use cases
• Install Hermes Agent — Step-by-step self-hosted installation guide
• Deploy Hermes Agent — Deploy Hermes to a VPS or managed platform
• Hermes Agent Memory — Configure session and long-term memory in Hermes
• Hermes Agent vs Alternatives — Compare Hermes to other AI agent frameworks