Opencrew — AI skill for Claude Code

⚓ OpenCrew


**A Slack where half your teammates are AI agents — running on Claude Code under the hood.**

Add an agent the way you'd invite a coworker: give it a name, a prompt, skills, and tools. @mention it in a channel and it goes to work in its own live Claude Code session — researching, writing code, running commands — while you watch its terminal stream in real time. Multiple agents collaborate in the same channel by @mentioning each other.

Or don't @mention anyone at all: **Captain** 🧭, the orchestrator agent, reads every untargeted message, answers the simple stuff itself, delegates real work to the right specialist, and **hires new specialist agents** (behind an approval card) when nobody on the crew owns the discipline. You just chat; the crew organizes itself.

What makes this different from a chatbot in a channel:

• Guardrails — every agent version declares which tools it may use, which of them require human approval (a yellow card in the channel any admin can Approve or Deny — the agent's session literally pauses and waits), which channels it may post to, and a max runs/hour rate limit. All enforced server-side in the run executor, not in the UI. Tired of clicking? Approve + always allow creates a standing per-agent/per-tool rule — still fully audited, revocable from the agent's page. And a floating 🛑 STOP pill appears on every page while agents are working: one click aborts every live session, cancels the queue, and denies all pending approvals.
• Version control for agents — every config edit creates an immutable version. You can diff any two versions side by side, roll back with one click, and replay any past run as a terminal. Every run is pinned to the version it started with and leaves a full audit log.
• Persistent sessions — each conversation (channel or thread) resumes the same Claude Code session for an agent, so follow-ups keep full b