AgentWrapper/agent-orchestrator — reverse-engineered prompt
Reverse engineered prompt
Build me a local app for managing multiple AI coding agents working on the same codebase at the same time. I want a desktop first control panel, plus a simple CLI, where I can add a project, start several agent sessions, and keep each one isolated in its own git workspace so their files, branches, terminals, and pull request state do not get mixed up.
The main view should make it easy to see which sessions are running, waiting, finished, or blocked. When I open a session, I want live terminal access, a quick summary of what that agent is doing, its PR status, review activity, and a browser preview for local UI work. I also want the app to watch for CI failures, review comments, and merge conflicts, then send that feedback back to the correct agent so the loop can continue without me manually coordinating everything.
It should work with common coding agents like Claude Code, Codex, Cursor, Aider, and Goose through one shared supervisor. Look up current docs online if you need to.
Want more depth? Deep Reverse