Docs

raunen is a terminal agent for local LLMs: one Go binary that reads and writes files, runs commands, and keeps the conversation where you can see it.

It works in the directory you start it in. Tools are rooted there, so raunen in a project is what gives it something to read. There is no separate window and no web UI — a running agent has nowhere else to put things.

Where to go

• Install — one command, or build from source with Go 1.25.
• Configuration — providers, models and where the files live.
• Modes — how much rope the agent gets, and how that is decided.
• Commands & keys — everything /help lists.
• Models & ladders — choosing one, and escalating when it runs out of room.
• Sub-agents — delegating investigation, several at a time.
• MCP servers — adding tools from outside.
• Local models — context windows, and the Ollama catch that bites everyone.
• Using tmux — jump between running instances with prefix + R.
• Wispr Flow — dictate prompts straight into raunen's input box.
• What it costs — free, and how to keep it that way.

The short version

Install it, have something serving an OpenAI-compatible API, and run raunen in a project. The first run writes a config, asks your endpoints what they serve and picks a model — preferring anything local. /help lists everything and ctrl+c leaves. The conversation is saved, so raunen --continue picks it back up tomorrow.