Launching today
Captain Barbossa
Your agent CLI, promoted to captain.
2 followers
Your agent CLI, promoted to captain.
2 followers
Captain Barbossa launches Claude Code, Codex, pi or Grok as a captain inside a Herdr workspace, then recruits further native agents as crew in new panes so a team works one checkout.
Hey Developers! 👋
I'm excited to share Captain Barbossa with you all!
I've been using AI coding agents like Claude Code, Codex, and Pi quite extensively, and one thing kept bothering me: why should I be limited to working with one agent at a time?
Sometimes I want one agent building a feature, another reviewing code, and a third investigating a bug. Managing all those sessions, keeping track of who's doing what, and coordinating their work quickly becomes a headache.
That's what led me to build Captain Barbossa. 🏴☠️
It's an open-source CLI that turns your favorite AI coding agent into a captain that can recruit and coordinate other agents as its crew.
The idea is simple:
- Use the coding agents you already love — Claude Code, Codex, Pi, or Grok.
- Delegate tasks to multiple agents working in parallel.
- Keep track of their progress, context, and token usage.
- Preserve useful memory across captain sessions.
- Work directly in your terminal, without another custom UI or orchestration server.
I wanted something that feels like a natural extension of my existing development workflow, not an entirely new environment to learn.
It's still evolving, and there are plenty of things I'd like to improve.
I'd especially love to hear from developers already using multiple coding agents: How are you coordinating them today? What's the biggest frustration in your workflow?
Give it a try, break things, open issues, or just share your thoughts. Feedback and contributions are always welcome!
GitHub: https://github.com/dev-preetamra...
Thanks for checking it out! 🚀