AI-Builder-Club/skills ? reverse-engineered prompt

Build me a Claude Code plugin marketplace project that gives me ready made skills for turning a repo into something agents can actually work in on their own, and for creating ongoing agent loops with a shared file based memory.

I want one plugin with two main entry points. One should set up a codebase harness so an agent can run the app, test it, verify its work, and ship changes safely. The other should create a new loop workspace with a shared knowledge base, basic docs like architecture and logs, and do an initial real run that writes back what happened.

Also include the supporting skills mentioned in the repo, like local dev setup, basic end to end test setup, optional isolated cloud boxes for parallel agents, a verifier that can capture proof like screenshots or video, and a utility that turns a workflow or article into a PNG and animated GIF flow diagram.

Keep it practical and usable in Claude Code, with clear install and usage flow. If you need details on current Claude Code plugin conventions, look up the docs online.