Ralph Orchestrator
Keeps a coding agent in a loop, with role 'hats', until the task is done
What it is
An implementation of the Ralph Wiggum loop technique: the orchestrator repeatedly runs a coding agent against a plan, switching between role presets it calls hats, until completion criteria are met. It is a Rust CLI with presets, an interactive planning step and a web dashboard, and supports several agent backends including Claude Code.
Good for
- Run an agent unattended on a planned feature
- Loop until tests and criteria pass
- Compare presets for different kinds of task
Install or run
npm install -g @ralph-orchestrator/ralph-cliCommand as given in the project README. Read the repository before you run anything on your machine.
What you need
Node.js or Cargo; at least one supported agent CLI
Sources
- https://github.com/mikeyobrien/ralph-orchestrator (project README, read 10 Nov 2026)
- https://api.github.com/repos/mikeyobrien/ralph-orchestrator (GitHub API, read 13 Nov 2026)
The summary is ours, written from the README. Spotted an error? See how we work.