Waleed Mujahid

Personal library of AI agent skills for Claude Code and other harnesses.

Gives Claude a free team of workers on Arbisoft's LiteLLM — including a browser. Interactive multi-turn workers (explore/debug/trace/exec/review/security/qa/playwright) Claude drives turn-by-turn; grunt work runs on free models, only findings re-enter Claude's context. Lighter, interactive successor to delegate.

Delegates grunt work to opencode running free LiteLLM models — saves Claude tokens. (Power-user alternative to pied-piper; pied-piper is the lighter, interactive default.)

End-to-end Sophia competency self-review assistant — phased chat that harvests a year of evidence, surfaces next-level rubric differentials, drafts rubric-aligned answers, and submits + verifies them via the Sophia API.