CLAUDE.md saturation curve
Project(s) touched: all (crescent's 98 commits / 60 days is the canonical signal); tooling fits github-io introspection
Status: Open — measurable but unmeasured
Surfaced in: docs/automated-introspection/investigations/2026-05-20-whats-wrong/synthesis.md, "What we still don't know" — "no measurement of whether commits-per-violation is rising (saturating) or falling (still mining)"
The question
Across the ecosystem, is CLAUDE.md churn approaching saturation (the rule set converging on the implicit-constraint set) or still mining (new rules per violation isn't falling)?
Why this is a registry thread
The measurement crosses every repo, and the tooling to compute it fits github-io's introspection layer. The answer tells us whether the rule-accretion process is converging or open-ended.
What's still open
- No metric for commits-per-violation over time.
- Needs a detector that joins CLAUDE.md commits to the violations that triggered them.
Working answer
None — unmeasured.