chekusu

chekusu / wanman

Public

wanman is an open-source agent matrix runtime inspired by Japanese one-man trains. It lets human users step back into an observer role while local agent runtimes coordinate autonomous multi-agent workflows, task execution, and artifacts.

wanman.ai agent claude-code codex llm
142
27
100% credibility
Found Apr 25, 2026 at 142 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

wanman is an open-source framework that runs teams of AI coding agents locally to autonomously analyze, plan, implement, test, and improve git repositories.

How It Works

1
💡 Discover wanman

You hear about wanman, a friendly team of AI helpers that can take over and improve your projects automatically.

2
📥 Get it ready

Download wanman to your computer and make sure your AI friends (like Claude) are logged in.

3
🚀 Hand over a project

Pick a folder with your code project and tell wanman to take over – it creates a safe copy to work on.

4
👀 Watch the magic

See your AI team chat, plan tasks, write code, run tests, and make pull requests right before your eyes.

5
📱 Chat and guide

Send messages to the team leader or check updates anytime to steer them toward your goals.

🎉 Enjoy improvements

Get back a better project with new features, fixed bugs, updated docs, and ready-to-merge changes.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 142 to 142 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is wanman?

wanman is a TypeScript CLI framework for running local multi-agent AI workflows, inspired by the Japanese wanman train where one operator handles everything autonomously. It spins up supervised networks of Claude Code or Codex agents on your machine, coordinating tasks, messages, and artifacts via simple commands like `wanman send`, `wanman task create`, or `wanman run `. Humans step back as observers while agents execute independently in isolated worktrees, producing deliverables like repo patches or structured outputs.

Why is it gaining traction?

Unlike cloud-heavy agent platforms, wanman keeps everything local and CLI-scriptable, making workflows reproducible and observable without vendor lock-in. The wanman train concept resonates—agents handle execution autonomously, with JSON-RPC for coordination and built-in tools for tasks, capsules, and hypotheses. At 142 stars, it's drawing devs who want woman ai setups that feel like a self-running train, complete with live `wanman watch` dashboards.

Who should use this?

Backend devs automating repo takeovers or goal-driven coding sprints, where you `wanman takeover ` and let agents like CEO/dev/feedback collaborate. Indie hackers building autonomous pipelines for artifacts and research, or teams experimenting with local agent matrices before scaling. Perfect if you're tired of manual AI prompting and want scriptable, human-back coordination.

Verdict

Grab it for local agent experiments—90% test coverage and multilingual docs (EN/JA/ZH) make it solid to tinker with, despite the 1.0% credibility score and modest stars signaling early maturity. Pair with Claude/Codex CLI for quick wins, but expect iteration on edge cases.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.