feat(ux): IA & navigation — judgments index, per-page titles, first-run fixes

[SoundMindsAI]

Fourth UX-audit fix PR. Scope: information architecture, navigation, and first-run content.

Changes

• Judgments index (highest-severity IA finding) — judgment lists were a core object with no list route, not in the nav, and every detail page shipped a "← All judgment lists" back-link that 404'd. Added /judgments (new JudgmentListsTable + column config) and a "Judgments" nav item. Filterless list (the useJudgmentLists hook is cursor/limit only). Fixes the 404 + the discoverability gap.
• Per-page titles — every tab previously read "RelyLoop". New useDocumentTitle hook applied to all list pages, plus a new DetailPageShell documentTitle prop wired into all 6 detail pages, so tabs/history/bookmarks read " · RelyLoop".
• Query-set → cluster traversal — the cluster on the query-set detail is now a link (was a raw, non-clickable UUID → a dead end).
• First-run copy — the dashboard subtitle now says what RelyLoop is (auto-tune relevance → ship a PR) instead of "Recent activity"; the chat example-prompt chips are capability-shaped ("Summarize one of my clusters") instead of naming demo entities (prod-es, trial 47) that don't exist on a fresh install.

Deferred (tracked follow-up)

Empty-state CTA parity (studies/query-sets/templates), cluster-tree breadcrumb consistency (F5), and reverse "used by" links (F7) — all low-severity; kept out to keep this PR reviewable.

Testing

useDocumentTitle + JudgmentListsTable (link + empty state) tests. The top-nav test iterates NAV_ITEMS, so the new item is auto-covered. Full suite green (1325); tsc + eslint + prettier clean (ran prettier across src).

🤖 Generated with Claude Code

[gemini-code-assist]

Code Review

This pull request introduces a new useDocumentTitle hook to dynamically set the browser document title across various pages and detail views (via DetailPageShell). It also adds a new "Judgments" page, including the JudgmentListsTable component, its column configuration, and associated tests. Additionally, navigation is updated to include Judgments, dashboard copy is improved, and chat example prompts are generalized. Feedback is provided regarding a potential hydration mismatch issue in Next.js when using toLocaleString() on a Date object in judgment-lists-table.column-config.tsx, suggesting the use of suppressHydrationWarning or client-side formatting.

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.

[gemini-code-assist]

Using toLocaleString() on a Date object during server-side pre-rendering in Next.js can lead to hydration mismatches if the server's locale or timezone differs from the client's browser settings. To prevent this, add suppressHydrationWarning to the wrapping element, or format the date only after the component has mounted on the client.

    cell: ({ row }) => (
      <span className="whitespace-nowrap" suppressHydrationWarning>
        {new Date(row.original.created_at).toLocaleString()}
      </span>
    ),

[SoundMindsAI]

Review adjudication

Gemini — 1 finding, rejected with counter-evidence:

Finding	Verdict	Reasoning
judgment-lists-table.column-config.tsx:55 — toLocaleString() may cause SSR hydration mismatch	Reject	This is a 'use client' table whose rows come from a client-side useQuery — the rows do not exist during SSR (the server renders the loading skeleton), so toLocaleString never runs server-side and no hydration mismatch is possible. The identical pattern is used across every existing list table (query-sets-table, etc.) without suppressHydrationWarning; adding it only here would be inconsistent cargo-culting. Centralizing date formatting (a real but separate polish item, visual-audit L4) is deferred to the theming/polish PR.

Only remaining red is the pre-existing backend (contract+integration) UBI test (#610). Merge-skew: clean.