Resolve Pr Feedback

This automates the tedious part of addressing PR feedback by spawning parallel agents to tackle each review thread independently, then marking them resolved. The philosophy here is refreshingly pragmatic: agent time is cheap, so fix everything valid including nitpicks rather than debating priority or adding tech debt. One security note worth flagging is that it treats comment text as untrusted input and won't blindly execute commands found in feedback, which is the right call. With 243 installs and some security audit warnings from Socket and Snyk, it's clearly being used in production but worth reviewing those findings before adding it to your workflow.