Show HN: I built a tool to assist AI agents to know when a PR is good to go
I've been using Claude Code heavily, and kept hitting the same issue: the agent would push changes, respond to reviews, wait for CI... but never really know when it was done. It would poll CI in loops...