OpenHermit User Manual

This is the user manual for OpenHermit. It is written for people who have an OpenHermit agent and want to use it — whether you run it alone or share it with a small team.

If you are looking for how to deploy, operate, or extend OpenHermit itself, that material lives in the rest of docs/ and is out of scope here.

Contents

Part 1 — Getting Started

  1. What OpenHermit Is
  2. Quickstart
  3. Core Concepts

Part 2 — Daily Use

  1. Talking to Your Agent
  2. Users and Identity
  3. Memory
  4. Instructions
  5. Skills
  6. MCP Servers
  7. Files and Workspace
  8. Schedules

Part 3 — Sharing Your Agent

  1. Inviting People
  2. Access Levels
  3. Managing Members
  4. Policy and Approval

Part 4 — Customizing (owner-only)

  1. Models
  2. Channels
  3. Secrets

Part 5 — Reference

  1. CLI Cheatsheet
  2. Web Admin UI Tour
  3. Troubleshooting
  4. FAQ
  5. Glossary

How to Read This Manual

  • First time with OpenHermit — read Part 1 in order (about 15 minutes). You will have an agent up and answering messages by the end.
  • Already using it, want to do something specific — jump to the chapter that matches. Each chapter ends with How-to recipes that follow a fixed shape: scenario → prerequisites → ways to do it → verify → common issues.
  • Planning to share your agent with others — Part 3 is written for you.
  • Stuck — start with Troubleshooting and the FAQ.

Conventions

  • Bold terms mark concepts on first appearance; each has a one-line definition in the Glossary.
  • Role markers: actions that only the agent owner can take are tagged (owner-only). The full capability table is in Chapter 5 · Users and Identity.
  • Command prefixes:
    • hermit … — CLI command
    • curl … — HTTP API call
    • Admin UI: path → action — Web admin operation