#251 — per-theme AA token repair + universal 8-theme AA matrix guard

[cmbays]

Summary

Repairs every (theme × token) pair the #264 adversarial verification measured below WCAG AA 4.5:1, at token level in templates/partials/tokens.css (the #206 deepening pattern — since #264 the tokens live in one place, so the report and explore pages repair simultaneously), and ships the universal 8-theme AA contrast-matrix headless guard so this class of regression is caught mechanically, not by audit.

Two surgical companions the guard pins:

• templates/report.css: .diff-sigil { opacity: 0.9 } removed — composited opacity is not used for text (the polish: latte suppress-chip 4.37 sub-AA + suppressed-row opacity vs AA (design decision) #227 rule); the per-theme sigil tokens now carry the full painted ratio, and the guard fails loudly if any self-or-ancestor opacity returns.
• templates/interaction.js: split-view .ds-num line-number cells gain aria-hidden="true" parity with the unified .diff-gutter (classified decorative per the design: shared block-diff palette + muted-text surfaces sub-AA on latte/rosepine/solarized (measured) #251 AC either/or — positional metadata duplicated from document order, the feature: design integration PR-2 — engine merge (settings panel, unified/split diffs) + port-forward of post-fork behaviors #178 idiom).

Latte/rosepine muted and the latte accent adopt the themes' own deeper steps already established by the #198/#227 scoped stand-ins (Catppuccin subtext1 #5c5f77, latte blue #1a5cdd, rosepine #5d5780). All other shifts are minimal lightness-only moves (hue/saturation kept). Light and Dark are untouched — they already passed everywhere.

Before → after AA matrix (measured live in headless Chromium, effective composited backdrops)

Theme	Surface	Before	After
solarized	accent nav link on --bg	3.41	5.00
solarized	diff add-sigil + on composited add-row	2.81	4.61
solarized	diff rem-sigil -	3.77	4.57
solarized	removed-line word-emphasis	3.53	4.59
solarized	muted (.code-filename / badge, worst-of)	4.33–4.50	4.75
latte	accent nav link on --bg	4.34	5.13
latte	accent-hover on --bg-alt	2.62	4.55
latte	muted on --bg / --surface-2	4.06–4.37	5.14–5.53
latte	removed-line word-emphasis	4.08	4.56
rosepine	muted on --bg (degrade/hint copy)	3.53	4.82
rosepine	row-count badge (muted on --surface-2)	3.77	5.15
rosepine	accent nav link on --bg	4.40	4.59
rosepine	accent-hover on --bg	2.47	4.56
rosepine	removed-line word-emphasis	3.80	4.58
tokyo	row-count badge (muted on --surface-2)	4.41	4.58
tokyo	added-line word-emphasis	3.27	4.60
tokyo	removed-line word-emphasis	4.22	4.61
gruvbox	row-count badge (muted on --surface-2)	4.17	4.59
gruvbox	diff rem-sigil	4.36	4.55
gruvbox	added-line word-emphasis	3.81	4.56
dracula	diff rem-sigil	4.47	4.56
dracula	added-line word-emphasis	4.05	4.64

Full post-repair sweep: 8 themes × 9 report surfaces = 72 cells, all ≥ 4.55 (plus the extended explore dag/tests sweeps).

The guard

report_aa_token_palette_matrix_on_every_theme (tests/headless_toggle.rs) renders one PR-diff report carrying every #251 surface and, per theme:

• measures effective composited colors — the backdrop walk collects every painted fill up to the first opaque one and alpha-composites top-down (the dark themes' rgba row/word tints are composited, never skipped);
• kills transitions/animations before any read; asserts checkVisibility() on every target (drawers opened first); pins the unified diff layout so the measured sigil/word instances are the painted ones;
• pins an opacity-dimming guard: any self-or-ancestor computed opacity < 1 on a measured surface fails loudly (the polish: latte suppress-chip 4.37 sub-AA + suppressed-row opacity vs AA (design decision) #227 rule);
• injects a bare <a> probe on <body> to exercise the shipped a { color: var(--accent) } chassis rule — the exact combo the C3 matrix measured on the explore nav;
• pins .ds-num aria-hidden parity with the unified .diff-gutter.

The explore twins (explore_dag_page_themes_correctly_on_every_theme, explore_tests_page_themes_correctly_on_every_theme) are extended with the C3 combos: muted-on-bg (counts/hint/test-shape), accent-nav-on-bg, and muted-on-surface-2 (the row-count badge — the matrix's worst muted offender, 3.77 on rosepine pre-repair).

RED-proof

With ONLY templates/partials/tokens.css reverted (git stash push -- templates/partials/tokens.css) and the guard code in place:

• the report guard FAILED with 24 sub-AA cells reproducing the measured C3 matrix verbatim — solarized/accent-link-on-bg = 3.41, solarized/diff-add-sigil = 2.81, rosepine/muted-hint-on-bg = 3.53, latte/row-count-badge-muted = 4.06, tokyo/diff-add-word = 3.27, gruvbox/diff-rem-sigil = 4.36, dracula/diff-add-word = 4.05, …
• the explore sweeps FAILED on the same pairs — [tests.html] rosepine/row-count-badge-muted-on-surface2 = 3.77, tokyo … = 4.41, gruvbox … = 4.17, [dag.html] solarized/nav-link-on-bg = 3.41, …

After git stash pop: all three guards PASS (worst cell 4.55). The guard demonstrably measures the painted, composited values it claims to.

Golden regeneration audit

All four example-report-check rows regenerated with the exact ci.yml recipes (jaffle-shop, playground, diff-showcase via report; explore via explore --out-dir), over the merged tree including #272's fixture changes. git diff --text -U0 -- examples/ audit: all 349 changed lines are token-value CSS, the #251 comment blocks, the sigil opacity removal, and the .ds-num aria-hidden parity — zero structural HTML drift. No root_path/username strings in any regenerated artifact (regenerated from committed synthetic fixtures only).

The render_integration insta snapshot was regenerated for the same inlined-CSS/JS changes (same audit, same result).

Gates (run directly, all green)

• cargo fmt --check ✓
• cargo clippy --all-targets --locked -- -D warnings ✓
• cargo nextest run — 1276 passed ✓
• cargo test --test bdd — 163 scenarios / 1044 steps passed ✓
• RUSTDOCFLAGS="-D warnings" cargo doc --no-deps --locked ✓
• cargo deny check ✓
• full headless suite --ignored: headless_toggle 72/72, headless_zero_egress 10/10 ✓

Closes #251

🤖 Generated with Claude Code

---

[coderabbitai]

Warning

Review limit reached

@cmbays, we couldn't start this review because you've reached your PR review rate limit.

More reviews will be available in 47 minutes and 55 seconds. Learn how PR review limits work.

Your organization has run out of usage credits. Purchase more credits in the billing tab to continue.

⌛ How to resolve this issue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans include higher PR review limits than trial, open-source, and free plans. In all cases, reviews become available again over time. During sustained high-volume PR review activity, CodeRabbit may temporarily slow when the next review becomes available.

Please see our Fair Usage Limits Policy for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 1b0ebeff-9b23-460f-b6c0-cdb628c4513c

📥 Commits

Reviewing files that changed from the base of the PR and between 17b1367 and efd0d50.

⛔ Files ignored due to path filters (1)

• tests/snapshots/render_integration__rendered_chrome_jaffle_shop.snap is excluded by !**/*.snap

📒 Files selected for processing (9)

• examples/diff-showcase-report.html
• examples/explore/dag.html
• examples/explore/tests.html
• examples/jaffle-shop-report.html
• examples/playground-report.html
• templates/interaction.js
• templates/partials/tokens.css
• templates/report.css
• tests/headless_toggle.rs

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch render-251-aa-token-palette

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[ghost]

Ready to review this PR? Stage has broken it down into 5 individual chapters for you:

	Title
1	Deepen theme tokens for WCAG AA compliance
2	Remove sigil opacity and add accessibility metadata
3	Implement universal AA contrast-matrix guard
4	Propagate theme repairs to example reports
5	Other changes

_{Chapters generated by Stage for commit efd0d50 on Jun 12, 2026 2:58am UTC.}

[gemini-code-assist]

Code Review

This pull request implements per-theme WCAG AA contrast ratio repairs (>= 4.5:1) across several themes (Tokyo, Solarized, Gruvbox, Latte, Rose Pine) by adjusting color tokens, removing opacity dimming on diff sigils, and adding accessibility attributes (aria-hidden="true") to split-view line numbers. It also introduces a comprehensive headless browser test sweep to verify these contrast ratios programmatically. The reviewer feedback suggests that when parsing the results of the headless contrast sweep, unexpected or malformed shapes should fail loudly using .expect() instead of being gracefully tolerated with unwrap_or('?').

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[github-actions]

📄 Rendered report preview

All golden examples regenerated cleanly.

🟡 Golden examples

Committed to examples/ and byte-identity gated — the canonical reports contributors and consumers browse. Stable across PRs.

Report	View	Download
playground-report.html	▶ Open ↗	⬇ Download
jaffle-shop-report.html	▶ Open ↗	⬇ Download
diff-showcase-report.html	▶ Open ↗	⬇ Download

🐶 Live dogfood preview

This PR doesn't touch dbt-project/, so there's no live dogfood preview.

▶ Open ↗ opens the report in your browser in one click —
published to this repo's GitHub Pages under /pr-273/.
⬇ Download fetches the same self-contained HTML as a workflow
artifact (auth-gated; works fully offline). Either way the report
makes zero external resource requests.

The Pages preview may take ~1 min to update after this comment
posts. On PRs from forks the Open link is unavailable (read-only
token) — use Download.

Alternative: GitHub CLI

# gh CLI >= 2.63 extracts into ./report-preview-playground/.
gh run download 27391567592 -R breezy-bays-labs/cute-dbt -n report-preview-playground
open report-preview-playground/playground-report.html

---

_{Posted by report-preview.yml for efd0d5053e5903f8873f4ac9a567ff90f810e8b6. Affordance only — never blocks merge.}

[cmbays]

Overnight orchestration wrap-up

Merged as a10964b (squash) under Christopher's overnight authority — CI fully green, push verified (origin == head efd0d50), 1 review thread dispositioned (Gemini fail-loud suggestion rejected with probe evidence: the unwrap_or(\"?\") is diagnostic-string-only; pass/fail is gated by the 8×9 cell-count assert + ratio > 0.0, a malformed fg/bg emits ratio −1 which trips loudly, and an .expect mid-loop would abort before the aggregated failure list is built — matching the assert_explore_page_themes precedent from #264). Issue #251 auto-closed ✔.

What shipped: 22 per-theme token repairs (solarized/latte/rosepine/tokyo/gruvbox/dracula), the #227 composited-opacity removal pinned, split-view aria-hidden parity, and the universal 8-theme × 9-surface AA matrix guard with alpha-composited effective backdrops. RED-proof reproduced the C3 matrix verbatim (24 failing cells with old tokens; worst cell 4.55 after repair).

Lineage note: built by a recovery agent after the original builder died on an API error — the WIP guard draft was intact but token repairs had never been written; they were derived fresh and verified against the C3 matrix.

🤖 Generated with Claude Code