Getting Started

Up and running in minutes.

Huginn works out of the box. Use managed mode for a fully self-contained setup with no external dependencies. Or point at an Ollama instance. Or bring your own API key for Anthropic, OpenRouter, or OpenAI. One binary, you choose the backend.

1

Install Huginn

One command. Drops a single binary into your PATH.

curl -fsSL https://huginn.sh/install.sh | sh

Or with Homebrew: brew install huginn

2

Choose your model backend

Pick one — you can change it later in ~/.huginn/config.json.

Managed (recommended — no external deps):

huginn runtime download

Ollama (if already installed):

ollama pull qwen2.5-coder:14b

Cloud API key (Anthropic, OpenRouter, OpenAI):

# Set provider + key in ~/.huginn/config.json
# See /docs/configuration for examples
3

Run huginn in any git repository

Navigate to a project and run huginn. It indexes your repo and drops you into the TUI.

cd ~/projects/my-app
huginn
4

Start building

Type a request. Use /plan to generate a reviewed plan before making changes, or /code to jump straight to implementation.

/plan add rate limiting to the authentication endpoint

What's next?

📖
Full Quickstart

Deeper walkthrough of your first session

 🧠
Named Agents

Create persistent AI personas that remember your codebase

 🎛️
Model Management

Download different models, switch between them

 ⚙️
Configuration

Tune Huginn to your workflow