Finding
The current token audit is total-only: 155 violations pass against a total baseline of 159, while category baselines moved in opposite directions. The 2026-09-06 measurement is raw-hex 61 and inline-svg 93, against the committed baseline raw-hex 45 and inline-svg 113.
Acceptance criteria
- Record independent per-rule ceilings or baselines for raw-hex, raw-rgb, raw-hsl, dark-prefix, hardcoded-shadow, and inline-svg.
- A decrease in one category must not finance a regression in another.
- Preserve the current measured baseline honestly; do not rewrite history to claim raw-hex 45 is current.
- Ratchet category debt down through focused fixes while keeping the global total informational.
- Add deterministic regression tests for category-level no-increase and staged debt migration.
This is later P2 maintenance work and must not displace #553 or R15.
Finding
The current token audit is total-only: 155 violations pass against a total baseline of 159, while category baselines moved in opposite directions. The 2026-09-06 measurement is raw-hex 61 and inline-svg 93, against the committed baseline raw-hex 45 and inline-svg 113.
Acceptance criteria
This is later P2 maintenance work and must not displace #553 or R15.