What problem does it solve? When two or more open pull requests address the same GitHub issue, maintainers must manually compare eligibility, correctness, and quality across each PR before deciding which to merge. This Skill automates that comparison with tiered gates, weighted scoring, and a deterministic verdict report. ## Core Features & Use Cases - Tiered evaluation pipeline: Runs six Tier 0 eligibility gates (open state, CI on PR SHA, mergeability, DCO and verified-commit compliance, branch protection, resolved automated-review threads), six Tier 1 correctness checks, and four Tier 2 quality checks with weighted scoring. - Supersession and attribution handling: Parses PR bodies for supersession statements, classifies transferred work, and enforces contributor attribution before recommending a replacement PR. - Deterministic verdict rendering: Validates the comparison spec and renders a markdown scorecard, behavior-coverage matrix, and merge or salvage recommendation via a Python renderer. - Use Case: An issue has three competing PRs. Run the comparator to discover candidates, gate them on CI and compliance, score their tests and quality, and receive a verdict naming the PR to merge or the one closest to ready. ## Quick Start Ask the agent to compare the open pull requests for a specific issue number in your repository and recommend which one to merge.
