guanyang/open-agent-hub — reverse-engineered prompt
Reverse engineered prompt
Build me a lightweight command line tool that acts like a hub for AI coding assistants. I want to be able to install it once, then from any project turn on or off reusable capabilities for tools like Claude Code, Cursor, Codex, Gemini, Trae, OpenCode, Kiro, and similar setups.
The main idea is that it should manage three kinds of things, skills, agents, and slash style commands, and link them into the right folders either for the current project or for my global user config. I should be able to list what is available, check status, enable everything, enable just one item by name, enable only skills or only agents or only commands, disable them again, choose a target assistant, or point it at a custom directory. It should also support syncing skills from configured upstream sources.
Please make it feel simple and clean, use markdown based definitions with metadata, and include sensible help text so it is easy to use from the terminal. Check current docs online if you need to.
Want more depth? Deep Reverse