ReviewDebt: a practical framework for scoring every pull request
summary
ReviewDebt introduces a deterministic score that computes review burden from five signal families — diff size and coupling, test-evidence delta, ownership spread, AI-authorship indicators, and evidence/rationale gaps — and outputs a CLI scanner report with reviewer minutes and concrete author actions. Examples: three side-by-side PRs (clean, high-debt, refactored) and a 90‑day scan over 524 PRs that sums to 228 senior reviewer hours and flags single-PR estimates up to 84 review hours; defaults, thresholds, and per-team weekly slopes are provided.
AI-written summary · about
chapters (18)
- 0:00Review debt impact
- 3:32PR metrics are vanity
- 4:59Defining review debt and its compounding effects
- 6:01Five signal families
- 7:28Signal one: Dev size and coupling
- 8:10Signal two: Test evidence gap
- 9:15Signal three: Directory spread
- 9:56Signal four: AI authorship
- 11:22Signal five: Evidence gaps
- 12:25Combining signals into score
- 13:27Demo: Low review debt PR
- 14:51Demo: High review debt PR
- 16:20Demo: AI authored PR
- 17:23Cross-repo scan results
- 19:56Actionable steps for review debt
- 21:20Five steps to adopt framework
- 22:47Why a number matters
- 23:51Key takeaways and next steps
related talks
47mGamifying Reduction of Technical Debt
45mCracking the Code Review
28mPull Requests: Merging good practices into your project
45mCracking the Code Review
38mCracking the Code Review: from Guesswork to Automated Guidance
28mHere Be Dragons: The Hidden Gems of Tech Debt
27mAsync non-blocking Code Reviews
11mRefactoring: Managing technical debt before it blows up in your face