Hermes Agent
Hermes Agent is an open source personal AI agent by Nous Research that runs in the terminal, a desktop app and messaging apps, and writes new skills for itself from completed tasks.
What Hermes Agent does
Hermes Agent is a general-purpose AI agent that runs on your machine and keeps working between conversations. You use it from a terminal interface, a desktop app, a web dashboard or messaging apps such as Telegram, Discord, Slack, WhatsApp, Signal and email, and all of them share the same memory and state.
Hermes Agent learns from its work: after a complex task it can write a new skill describing how it did it, and a built-in curator maintains those skills over time. It also has a cron scheduler that delivers results to any connected platform, isolated subagents, and seven terminal backends, from local and Docker to SSH, Modal and Daytona.
Hermes Agent works with Nous Portal, OpenRouter, OpenAI, Anthropic, Google, a custom endpoint, or self-hosted Ollama, vLLM and LM Studio. The agent itself is free under MIT; Nous Research sells optional Nous Portal plans for hosted models, from a free tier to paid monthly plans.
Who Hermes Agent is for
- People who want an always-on assistant reachable from messaging apps and a terminal
- Recurring jobs that should run on a schedule and report to a chat
- Users moving from OpenClaw or Claude Code who want to import their setup
- Running an agent on local models or your own server
Who should pick something else
- Developers who want a coding agent inside an editor; Cline, Cursor or Zed fit better
- Anyone who wants a single packaged app with no configuration; setup involves providers, backends and config.yaml
- Users on Linux who want the desktop app: Linux desktop packaging is disabled in the release workflow per the platform docs
Hermes Agent: one strength, one limit
Strength. Hermes Agent is MIT licensed with 250,486 GitHub stars (2026-10-01) and turns completed tasks into reusable skills on its own, stored in ~/.hermes/skills.
Limit. Hermes Agent skills and memory live in its own ~/.hermes folders, so project skills need an explicit hermes skills trust step before they load.
What Hermes Agent costs
Free and open source under MIT with your own provider key or a local model; Nous Research also sells optional Nous Portal plans for hosted models.
Current prices: hermes-agent.nousresearch.com . We do not list prices: vendors change them often, so check the maker's own page before you buy.
Hermes Agent with agent skills and MCP servers
Skills. Hermes Agent has its own skills system, compatible with the agentskills.io standard. Bundled, installed and self-written skills live in ~/.hermes/skills; project skills in .hermes/skills or .agents/skills load only after hermes skills trust, and extra folders such as ~/.agents/skills can be added under skills.external_dirs in ~/.hermes/config.yaml. Every skill becomes a /skill-name command. The open skills CLI 1.7.0 installs Hermes skills into .hermes/skills. The docs give no date skills were added.
MCP. Hermes Agent reads MCP servers from ~/.hermes/config.yaml under the mcp_servers key: command and args for a local server, or url and headers for a remote one. hermes mcp opens an interactive picker, and hermes mcp install <name> adds a server from a catalog Nous reviews. hermes import-agent claude-code copies MCP servers, skills and instructions from a Claude Code setup.
Browse agent skills and MCP servers that work with Hermes Agent.
Alternatives to Hermes Agent
- OpenClaw: OpenClaw is an open source personal AI agent from the OpenClaw Foundation that runs on your own machine and takes tasks from WhatsApp, Telegram, Slack, Discord and other chat apps.
- Goose: Goose is an open source AI agent, created by Block and now part of the Agentic AI Foundation at the Linux Foundation, that runs as a desktop app and CLI with any model provider.
- Claude Code: Claude Code is a terminal coding agent by Anthropic that reads a codebase, edits files and runs commands, and also works in VS Code, JetBrains, a desktop app and the web.
- OpenCode: OpenCode is an open source terminal coding agent by Anomaly that works with 75+ model providers, local models included, and also ships as a desktop app and IDE extension.