Show HN: Built a small CLI for self-improving OpenClaw agent loops

1 points by j0xnvm 21 days ago · 0 comments · 1 min read

It reviews a plan or output with 1–3 LLMs and returns: PASS / REVISE / BLOCK

Use case:

- review a plan before execution - review an output before sending or relying on it - loop until the result is good enough to proceed

Very small repo (255 lines of code)

No comments yet.