What problem does it solve? Keeping a project backlog accurate is hard: completed stories may lack tests, open issues may miss priorities or dependencies, and milestones can drift from the actual architecture. This Skill provides a structured protocol for adversarial backlog grooming that verifies issues against the real codebase and files new stories for discovered gaps. ## Core Features & Use Cases - Codebase Ground-Truth Verification: Compares open and closed GitHub issues against actual code in apps, packages, migrations, and integration tests to confirm claimed work truly exists. - Adversarial Gap Detection: Systematically hunts for failure modes such as webhook replay attacks, missing DB transactions, race conditions during drops, and secrets leakage. - Story Refinement & Creation: Enriches issues with file paths, acceptance criteria, dependency links, and mandatory priority labels, and files new issues via the GitHub CLI. - Roadmap Auditing: Runs automated audits that flag legacy architecture keywords in milestones, verify deliverables on disk, and enforce priority-label compliance. - Use Case: Before a sprint planning session, run the refinement workflow to verify recently closed stories, detect unblocked shovel-ready issues, and produce a grooming report with escalated decisions. ## Quick Start Ask the AI to run an adversarial backlog refinement pass over the repository's open GitHub issues and produce a grooming report with refined stories and gaps.
