RuFlow: Run 60+ AI Agents in Claude Code (Free Setup)
What RuFlow is
RuFlow is a free, open-source repo that turns one Claude Code session into a swarm of sixty-plus AI agents that build what you ask for and get sharper after each run. It was built by ruvnet under an MIT license and has sat at the top of GitHub, yet most people have never set it up. An agent is a worker that goes off and finishes one job on its own, so sixty agents means sixty jobs moving at once instead of you stepping through them one at a time.
What is happening under the hood
Queen agents run the teams, and worker agents under them do the real work: the coding, the testing, everything in between. They share one memory, so the swarm reuses what worked on the last run. The part that saves money is the routing. RuFlow reads how heavy each job is and sends it to the right model on its own, so light work goes to a cheap model and heavy work goes to a powerful one. That cuts token usage by roughly half and stretches your Claude Code usage about 250% without you changing how you work.
Before you start
You need Node.js installed and Claude Code already running, since RuFlow sits on top of Claude Code instead of replacing it.
npm install -g @anthropic-ai/claude-code
claudeStep 1 — Initialize RuFlow in your project
Open your terminal in your project folder and run the wizard, which walks you through setup:
npx ruflo@latest init wizardThat installs the agents, their shared memory, and the routing layer that decides which model each job goes to. You run it once per project. To keep RuFlow inside Claude Code as a plugin instead:
/plugin marketplace add ruvnet/ruflo
/plugin install ruflo-core@ruflo
/plugin install ruflo-swarm@rufloStep 2 — Talk to Claude Code like you always do
Once it is initialized, use Claude Code the normal way. RuFlow's hooks sit underneath, routing each task and spawning the agents it needs. Give it a real target:
Build a REST API for a notes app with user auth, CRUD endpoints, and tests. Research the structure first, then have worker agents write the code and a reviewer agent check it.The queen agent splits that across the workers and brings back the finished build.
Step 3 — Let the shared memory compound
On every run the swarm saves what worked and reuses it, so it gets faster and more accurate the longer you keep it on a project. Your second feature tends to come out cleaner than your first.
Frequently asked questions
- Is RuFlow free? — Yes. It is open-source under the MIT license, with nothing to subscribe to.
- Does it replace Claude Code? — No. It sits on top of Claude Code, so you keep working the way you already do.
- How much does RuFlow save? — Around half the token usage and about 250% more runway, from the automatic model routing.
- What do I need to run it? — Node.js and a working Claude Code setup.
Get RuFlow
The repo, full docs, and the complete agent list live here: https://github.com/ruvnet/ruflo