Engineering Quality measures whether baseline engineering and documentation practices are in place: continuous integration, automated tests, lint and formatting discipline, and documentation a newcomer can actually learn from. These are the practices that make change safe — for human contributors and, increasingly, for AI coding agents.
The category carries 20% of the overall health index.
Metrics in this category
| Metric | Weight in category | Weight in overall |
|---|---|---|
| Engineering practices | 60% | 12% |
| Documentation | 40% | 8% |
Engineering practices is a weighted checklist of hygiene evidence: CI workflows, a test suite, linter configuration, pre-commit hooks, and an .editorconfig. Documentation measures whether the project explains itself: README, a documentation directory or site, repository description, topics, and wiki.
What this category deliberately measures
The signals are presence-based: they confirm the infrastructure exists, not how good it is. That restraint is intentional — grading test quality or documentation prose from the outside would be guesswork presented as measurement. Presence signals are cheap to verify, hard to argue with, and already separate disciplined projects from undisciplined ones remarkably well. Deeper signals (coverage, CI pass rates) are on the methodology roadmap and will arrive with a version bump.
Reading the category value
- A low value here rarely lies: missing CI and missing tests are visible facts, not judgement calls.
- A high value means the scaffolding of quality is present. It does not certify that the code inside is good — see signals, not warranties.
- Documentation weighs 40%: a well-tested project that explains nothing to newcomers is measurably less adoptable, and the category reflects that.
Improving Engineering Quality
- Add CI that runs the test suite on every push and pull request — the two heaviest components in engineering practices.
- Commit linter and formatter configuration; add pre-commit hooks and an
.editorconfigfor full marks. - Maintain a real README, a
docs/directory or documentation site, a repository description, and topics — each is an individually weighted component of documentation.
Related: AI readiness · the health index