feat: roll out Batch 2 product intelligence visuals

[ericsocrat]

Purpose

Roll the approved Folded Label Register and Evidence Register visual language through Batch 2 product-intelligence routes without changing scoring, provenance semantics, mutation behavior, database policy, dependencies, baselines, or performance methodology.

Behavior changes

• Redesign product detail, detailed analysis, nutrition, evidence, alternatives, compare, saved comparisons, watchlist, lists, and list detail.
• Reuse and extend ProductRegisterCard with evidence-aware score semantics and native meter output.
• Keep missing, provisional, stale, uncollected, loading, and unavailable states distinct.
• Withhold recommendations and comparison winners when source/freshness evidence is invalid.
• Remove list-preview request fan-out and partial average-score claims.
• Preserve application-level failure handling and optimistic rollback contracts.
• Replace the More drawer backdrop click reliability issue with explicit dialog keyboard handling while preserving native modal, cancel, focus, swipe, and restoration behavior.
• Fix the mobile navigation label-in-name defect found by the focused axe run.

Verification

Passed locally:

• TypeScript type-check
• ESLint
• production Next.js build
• 6,889 of 6,891 executed full-suite tests; 19 skipped
• 134 focused shared product/Search/Categories/Watchlist tests
• 256 design-system tests
• deterministic live inventory regeneration
• guarded local-authenticated Chromium Batch 2 journey
• 37 screenshots across 390, 768, and 1440 in light/dark plus forced colors
• Polish and German overflow checks
• focused product-detail mobile axe audit
• tracked-source security hygiene scan
• immutable baseline and dependency diffs are empty

Local environment-only failures:

• non-production Manrope license byte assertion sees CRLF-expanded Windows checkout bytes
• base-owned intentional-redesign mutation runner exceeds its local 15-second timeout

Accessibility

• Explicit loading/error announcements and retry actions remain available.
• Native score meters replace generic meter roles.
• Product tabs, keyboard focus, mobile navigation naming, forced colors, reduced motion, and 44px primary targets were checked.
• Focused mobile product-detail axe audit passes.

Performance

• No Lighthouse run or performance-policy change.
• No dependency or lockfile change.
• Route-JS remains owned by normal CI.

Visual evidence

The guarded run produced transient review screenshots for Product summary, analysis, alternatives, Compare, Watchlist, and Lists at 390x844, 768x1024, and 1440x900 in light/dark, plus a forced-colors product state. Baseline files remain byte-identical.

Merge hold

Keep this PR draft and unmerged until:

• Sonar issue AaBDzthQjq04_D568HgF is classified FALSE_POSITIVE with the approved rationale using authenticated Sonar access.
• Sonar confirms the More drawer reliability issue AaBGDLHCPm4zC770525e is resolved.
• Required Linux CI, immutable comparator, Route-JS, CodeQL, Quality Gate, Golden Reference, and normal deployment checks finish.

The Sonar false-positive control was unavailable locally because both safe browser sessions were signed out. No quality profile, threshold, Auth code, or provider configuration was changed.

Not run

• Lighthouse
• broad manual Firefox/WebKit matrix
• baseline generation or acceptance

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
tryvit	Ready	Preview	Aug 28, 2026 9:24am

[ericsocrat]

Batch 2 exact-head certification: HOLD MERGE

• Base: 1763c1e
• Head: 0921684
• Head tree: 7e859e04e81c2a27fdbc57b838e32eceee5f944d
• Synthetic merge: 7926d81
• Synthetic tree: 7e859e04e81c2a27fdbc57b838e32eceee5f944d

Product-relevant gates PASS: PR Gate, CodeQL, Quality Gate, Golden Reference, Route-JS, immutable comparator, intentional-redesign validation, repo hygiene, screenshots, and Vercel preview. Open code-scanning alerts: 0. Open secret-scanning alerts: 0. Unresolved review threads: 0.

Lighthouse was intentionally cancelled during dependency installation. Both measurement steps were skipped; no new samples were produced.

Inherited renderer attestation remains failed on exact base-owned policy tests and is not caused by this PR.

Sonar merge hold:

• AaBGDLHCPm4zC770525e remains OPEN on main; this PR contains the bounded More drawer fix, but Sonar has no PR 1324 analysis.
• AaBDzthQjq04_D568HgF remains OPEN and requires the approved FALSE_POSITIVE classification with authenticated Sonar access.

Keep this PR draft and unmerged until those Sonar dispositions are completed.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 092168418d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Keep navigation badge counts accessible

When listCount or compareCount is positive, aria-hidden removes the only count text from the accessibility tree, while this change also deletes the count-bearing aria-label; screen readers therefore announce only “Lists” or “More” and cannot discover the badge’s list or selected-product count. Preserve the visible-label-in-name fix with label-first accessible text such as “Lists, 3 items” or an equivalent visually hidden description.

Useful? React with 👍 / 👎.