bmad-os-findings-triage

Human-in-the-loop triage of code review findings works by parsing severity and file paths from reports, creating one task and one agent per finding, and tracking readiness to gather human decisions in an organized batch workflow.

The best way to manage a batch of PR review findings is orchestrating parallel agents that extract titles and severity, track completion status, and handle defer actions and cleanup for the final wrap-up.

This finding triage workflow supports prompt audit reports by parsing inline notes or pre-populated task lists, dispatching dedicated agents per finding, and maintaining organized readiness tracking until human review is complete.

You track readiness and decisions for multiple review findings by using a workflow that creates one task and one agent per finding, monitors their completion status, and records the final human decisions during batch wrap-up.

You can defer or shutdown agent orchestration during finding triage, as the workflow explicitly handles batching, defer actions, shutdown procedures, and cleanup to keep review decisions organized without losing track.