getcrew44

getcrew44 / crew44

Public

Orchestrate a crew of specialist AI agents in one local-first workspace. Each role on its best model, with memory and skills that compound. Free, MIT.

crew44.io ai-agents claude-code codex coding-agent developer-tools
34
1
89% credibility
Found May 27, 2026 at 34 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Go
AI Summary

Crew44 is a desktop application that coordinates multiple AI coding assistants on your computer into a working team. Instead of using one AI assistant alone, you can assign different tasks to specialists — one agent plans, another codes, a third reviews — and they hand work off to each other automatically. Everything runs locally on your machine with no cloud account needed. The app discovers AI tools you already have installed, lets you attach reusable 'skills' that improve over time, and keeps your project knowledge in memory across tasks. Think of it as a project manager for your AI agents.

How It Works

1
💬 You tell your assistant what you need

Open Crew44 and type a request like 'build me a login page that handles 401 errors gracefully'.

2
🤖 Your specialist team gets to work

Instead of one agent, a Planner drafts the approach while an Engineer writes the code and a Reviewer checks the work — they hand off to each other automatically.

3
🧠 The team remembers your project

Anything important about your project — like 'we use pnpm, never npm' — gets saved as a memory that every agent sees on every task.

4
⚡ You connect your favorite AI tools

Point Crew44 at the AI assistants you already have installed — Claude Code, Codex, Cursor, Gemini, and more — and it discovers them automatically.

5
🔒 Everything stays on your computer

No accounts, no cloud, no sharing. The only data that leaves your machine is what your chosen AI assistant already sends to its own service.

✨ Your project comes to life — together

Multiple specialists work in parallel on complex tasks, handing off seamlessly, and the results are exactly what you asked for.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 34 to 34 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 crew44?

Crew44 is a desktop application that turns your existing AI coding tools into a coordinated team. Instead of running one generalist agent, you assemble specialists—each bound to the model that does its job best—and let them hand work off to each other. The backend is written in Go, the UI is Electron with React 19, and everything runs locally on your machine. State lives in plain files under `~/.crew44/`, no cloud account required. It supports Claude Code, Codex, Cursor Agent, Gemini CLI, and several other agent runtimes, routing tasks to whichever CLI you have installed.

Why is it gaining traction?

The pitch is compelling: skills that compound. Write a workflow once as a `SKILL.md` file and every agent you attach it to gets it on every turn. Memory persists across sessions, so Monday's fix is in the crew's muscle memory by Thursday. The default crew ships with a Partner, Engineer, Product Lead, and Designer, each with distinct instructions and skills. There's also an auto-optimizer that scans your chat history and writes new skills back to disk. For teams already paying for multiple AI subscriptions, running specialists on the right model per role—without adding yet another subscription—is the hook.

Who should use this?

Developers who already use Claude Code, Codex, or Cursor and want to scale beyond a single-agent workflow. Product teams who want AI assistance that remembers project conventions. Shops running multiple AI providers and tired of context-switching. Not ready for teams needing enterprise support or SLA guarantees—34 stars means this is early-stage software.

Verdict

Crew44 solves a real problem with a clean architecture, but at 34 stars it's still finding its footing. The MIT license and local-first design are right for developers who value privacy and control. With a credibility score of 0.8999999761581421%, the codebase shows intentional engineering—recovery from crashes, scoped skill injection, end-to-end encrypted mobile pairing—but maturity takes time. Worth watching; maybe worth trying on a side project.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.