PotapKong/after-ai-forge — reverse-engineered prompt

Build me a starter kit that turns Claude Code into a small AI dev team with memory. I want one main orchestrator and six specialist agents for UI design, frontend, backend, testing, security review, and final verification. The agents should hand work to each other through project docs, ask for requirements before building, use test driven work for frontend and backend, and write down lessons so the system gets better over time.

Please include a persistent memory service using Postgres with pgvector, a simple command line tool to recall and save memories, Docker setup, and clear install instructions for Linux or WSL. I also want a workspace template for new projects, with folders for design specs, API contracts, test reports, security audits, verification results, and lessons learned.

Make it easy to copy the agent definitions into Claude Code, start the memory service, check health, and understand the full workflow. Look up current docs online if you need to.