Documentation
Everything you need to get Huginn running and make the most of it.
Getting Started
Core Concepts
- How Huginn Works — the agentic loop and write approval
- Named Agents — persistent AI teammates
- Sessions — history, memory, and context compaction
- Models — local and cloud providers
Features
- Slash Commands — full command reference
- Permissions & Write Approval — staying in control
- Spaces — channels and DMs for multi-agent work
- Skills — inject rules into every agent
- Built-in Tools — file ops, git, shell, search
- MCP Integration — connect any MCP server
- Agent Toolbelt — control what each agent can access