Tracking repository health at scale
summary
Trainline defines repository health programmatically, running checks (node LTS, dependency drift within two major versions, GitHub permissions, security audit, use of design system, banned packages like momentjs) across an estate that grows to 160 repositories and stores a daily JSON snapshot per repo for 18 months. Scores use pass/warn/fail/indeterminate, simple percent and rag thresholds (40/60). Results publish as a GitHub Pages dashboard, alert Slack on regressions, feed PlanDeck, and block or thank PRs as enforcement options.
AI-written summary · about
chapters (10)
- 0:00Repository health tracking intro
- 1:05Why track health at scale
- 6:18Repository health definition
- 8:21Key health checks
- 13:19Health scores and thresholds
- 15:45Archiving low-value repos
- 17:56Health dashboard explained
- 20:44Health checks for regressions
- 22:51Gamifying and PR integration
- 24:14Conclusion and key takeaways
related talks
33mKeep your dependencies in check
58mProtect your code with GitHub security features
23mRelease Status Analyser to bring more agility @ work
31mGitHub as a self-sustaining product for large-scale environments
43mTop Ten Worst Repositories to host on GitHub
05mContinuously Update Everything
31mIf You Didn't Record It It Didn't Happen
08mAll eyes on Open Source: Monitoring community sovereignty