feat(filters): therapy-compass adopts the filter contract (PR E)

[BigSimmo]

Summary

• Converges therapy-compass's bespoke phone-only TherapyFilterSheet / TherapyFilterTrigger (filter-sheet.tsx, now deleted) onto the shared ResultFilterSheet / ResultFilterTrigger / ResultFilterFacetChips (docs/filter-contract.md), fixing both invariants the contract doc already named against this mode: the sheet's onClearAll no longer wipes the search box (the composer's own "Clear search" already covers the query on every breakpoint, matching every other adopted mode), and the trigger is the shared ResultFilterTrigger instead of a bespoke reimplementation with a different icon, a hardcoded test id, and a different label-hiding breakpoint.
• Three facet groups, one array shared by the desktop rail and the phone sheet: Topics (the six curated quick-filter tags) is many-of-N, OR within the group — this also fixed searchTherapies' tag matching, which had been AND-within-group (a therapy needed every selected tag). That's the same defect class already fixed for document tags: a second topic selection narrowed instead of widening, making multi-select within one group a dead affordance instead of a feature. Review status and Handout (Reviewed only / Brief available) are modelled as two independent one-option facet groups rather than options inside one shared group, because searchTherapies ANDs them against each other and against Topics — a shared group would OR them, which is not what the real filter does.
• matchesTopics/matchesAvailability in data/select.ts are the single predicate both searchTherapies (the real filter) and the sheet's option counts (option.hint) run through, so a count can never drift from what clicking the option actually does (docs/filter-contract.md section 3).
• Desktop rail: three ResultFilterFacetChips groups render side by side; neutralised the shared component's border-t divider (meant for groups stacked vertically inside the sheet) with a [&>section]:border-t-0 wrapper so it doesn't show as a floating hairline above the second/third group in this new side-by-side layout — verified in a real browser before and after.
• Updated docs/filter-contract.md (section 6's two invariant callouts, plus a new Rollout entry) and tests/therapy-compass-mode-wiring.test.ts's static wiring assertions, which had locked in the old bespoke sheet's deliberate full-reset onClear={b.clearSearch} behavior — now updated to assert the new onClearAll contract instead.

Verification

• npm run verify:pr-local — all 15 steps green (build run with the local dev server stopped, per the gate's own refusal check)
• npm run test — full unit suite, 565 files / 6167 passed / 4 skipped / 0 failed
• npm run typecheck — clean
• npm run lint — clean
• npm run check:bundle-budget on a rm -rf .next fresh build — production 1304.4 KiB vs 1279.1 KiB baseline, mockups 294.5 KiB vs 267.5 KiB baseline, both within tolerance
• Browser proof (Playwright/Chromium) at 320/390/639/768/1440/1920px — 0px horizontal overflow at every width; phone filter trigger measures 48×48px (exceeds the 44px floor its own accessibility test pins); live-verified widen-on-multi-select counts (e.g. selecting CBT then DBT both show the union count, not two independently-narrowed numbers) and that Clear all resets every filter while leaving the query text visibly intact in the results header behind the sheet
• npm run verify:ui not run — this environment's Playwright is pinned to a browser revision (r1234) not preinstalled here (r1194); direct Chromium browser proof above substitutes for the routes this PR touches. Full Chromium gate is a normal CI/hosted-environment run.

Risk and rollout

• Risk: Low. Scoped to therapy-compass's own search screen and its local (non-Supabase, non-RAG) data module; no other mode's filter sheet, trigger, or facet renderer is touched. The Topics AND→OR behavior change is a deliberate, documented fix (matching the already-established convention every other adopted facet in the app uses), not an accidental regression — flagged explicitly here rather than left implicit.
• Rollback: Revert this commit; filter-sheet.tsx and its test file are recoverable from history if a revert is ever needed.
• Provider or production effects: None. No Supabase, OpenAI, or other provider surface touched; the therapy catalogue is a local generated fixture.

Notes

• Real local fixture data note (not a defect): every therapy in the current 205-item catalogue has reviewStatus: "needs_review" — none has cleared review yet — so the "Reviewed only" facet chip currently shows (0) regardless of other selections. It is intentionally not hidden or disabled: it's a legitimate, functioning filter whose count reflects real (if currently empty) data state, exactly matching how the old bespoke toggle behaved before this PR (always shown, unconditionally). It will show real counts once therapies clear review.
• This is PR E of the filter-contract rollout (docs/filter-contract.md). Prior PRs in the same rollout (lens modes, formulation, services, factsheets) are separate branches/PRs at various stages — this PR is independent and does not depend on any of them merging first.

---

Generated by Claude Code

[supabase]

This pull request has been ignored for the connected project sjrfecxgysukkwxsowpy because there are no changes detected in supabase directory. You can change this behaviour in Project Integrations Settings ↗︎.

---

Preview Branches by Supabase.
Learn more about Supabase Branching ↗︎.

[coderabbitai]

Warning

Review limit reached

You’ve reached a temporary PR review limit under our Fair Usage Limits Policy.

Your recent review volume is higher than typical usage, so adaptive limits are currently applied.

Next review available in:2 minutes

Your organization has reached its usage spending cap. Adjust your spending cap in the billing tab.

How can I continue?

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.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 587f8229-d18a-439f-9dcb-43cce29c9cf4

📥 Commits

Reviewing files that changed from the base of the PR and between 2ae5a0a and 5c165f2.

📒 Files selected for processing (12)

• docs/design-system/COMPONENTS.md
• docs/design-system/adoption-manifest.json
• docs/filter-contract.md
• src/components/therapy-compass/bindings.tsx
• src/components/therapy-compass/data/select.ts
• src/components/therapy-compass/filter-sheet.tsx
• src/components/therapy-compass/screens/search-screen.tsx
• tests/therapy-compass-mode-wiring.test.ts
• tests/therapy-compass-search-filters.dom.test.tsx
• tests/therapy-filter-sheet.dom.test.tsx
• tests/therapy-search-facets.test.ts
• tests/ui-accessibility.spec.ts

---

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

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit:18e2b7ce97

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".

[BigSimmo]

@codex resolve actionable Codex review findings for this pull request and current head using the repository instructions. This is the pull request's single automatic repair pass: do not perform a fresh review, create new standalone findings, or request another review. Work only the existing unresolved Codex threads on the current head. The only repository destination is BigSimmo/Database, and the only branch destination is the pull request head branch claude/filter-therapy-compass at starting commit 0bbe967; never publish fixes to a detached or synthetic work branch and never create a stacked pull request. Use the authenticated GitHub connector to commit each approved fix to BigSimmo/Database:claude/filter-therapy-compass, then verify that the pull request head contains the published commit before reporting success. Always fix P0 and P1 findings. For P2 and lower findings, fix only clear, scoped, low-risk issues; otherwise disposition them with a concise reason. For a fixed thread, reply with as the first line and as the second line. For a no-code disposition, use followed by . These result markers authorize the workflow to close that exact thread only after it verifies a fixed commit is the pull request head; a local-only commit is not a fix. If publication or verification fails, do not use either result marker, do not claim success, and leave the thread open with the blocker. If human input or new authorization is required, do the same. Finish only after every actionable thread is fixed or dispositioned and closed, or explicitly left open for a human decision. Do not update the branch from main, address unrelated reviews, broaden scope, or create more than one scoped fix commit. Do not use external APIs, paid services, credentials, dependency changes, or broad refactors unless explicitly authorized. Add targeted tests where behavior changes and run the narrowest relevant validation.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. What shall we delve into next?

Reviewed commit:0bbe967f40

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".

[BigSimmo]

Addressed the review finding on the current head 6d57b773012070f320198f35840c221f7bb57c5c: unselected zero-result topic, reviewed-only, and brief-available options are now exposed as focusable aria-disabled dead ends and guarded from toggling.

[BigSimmo]

@claude Work only the current unresolved review finding on this PR at head 6990ce6eb01b8bf96093939d6343cd36c1ab081c. Verify the P2 in src/components/therapy-compass/screens/search-screen.tsx: unselected facet options with an honest zero result count remain actionable. If still valid, minimally mark zero-count unselected Topic and Availability options disabled while preserving already-selected zero-count options so users can remove them, add focused DOM coverage proving aria-disabled/untoggleable behaviour, run the narrowest format/type/test checks, push the fix to this PR branch, reply with the pushed SHA and verification, and resolve the satisfied thread. Do not merge, close, rebase, force-push, deploy, alter auto-merge, or broaden scope.

[github-actions]

Hi @BigSimmo — thanks for the @claude mention.

Open this in Claude Code on the web to continue here: https://claude.ai/code

Context: BigSimmo/Database · #1889

_{Automated backlink from the Claude backlink workflow. Claude Code can pick up the mention and push a scoped fix to the branch.}

[claude]

Claude encountered an error after 2s —— View job

---

I'll analyze this and get back to you.

[BigSimmo]

@claude The current review fix is already on head 6d57b773012070f320198f35840c221f7bb57c5c and its thread is resolved, but the PR is now mergeable_state: dirty against current main. Fetch current main and merge it into this PR branch without rebasing or force-pushing. Resolve only merge conflicts, preserving the verified zero-count facet-disable fix and all newer main changes. Run conflict-adjacent format/typecheck plus focused therapy-compass filter tests, push the merge-resolution commit, and report the pushed SHA and verification. Do not merge or close the PR, deploy, alter auto-merge, or broaden scope.

[github-actions]

Hi @BigSimmo — thanks for the @claude mention.

Open this in Claude Code on the web to continue here: https://claude.ai/code

Context: BigSimmo/Database · #1889

_{Automated backlink from the Claude backlink workflow. Claude Code can pick up the mention and push a scoped fix to the branch.}

[claude]

Claude encountered an error after 2s —— View job

---

I'll analyze this and get back to you.

[github-actions]

CI triage

CI failed on this PR. Automated classification of the 4 failed job(s):

• Static PR checks — needs investigation: inspect the failing step and uploaded diagnostics; rerun only after classifying the cause.
• Unit coverage — needs investigation: inspect the failing step and uploaded diagnostics; rerun only after classifying the cause.
• Production UI (1) — needs investigation: inspect the failing step and uploaded diagnostics; rerun only after classifying the cause.
• PR required — needs investigation: inspect the failing step and uploaded diagnostics; rerun only after classifying the cause.

Compared with main CI run #10317 (success).

Classification is evidence routing, not permission to ignore a failure. Exact quarantined Playwright identities remain governed by the flake ledger.

[BigSimmo]

Updated the filter DOM fixture so the independent Reviewed-only toggle is exercised with a reviewed therapy, while preserving the zero-count disabled guard. The PR remains blocked by its base-branch merge conflict; no rebase or force-push was performed.

[BigSimmo]

@claude Please main-sync this PR from exact head 4caf4e690cd8b9704a442683943e08165962772b: merge the latest main once (never rebase), resolve every conflict minimally, commit and push. Do not merge/close the PR, force-push, alter auto-merge, or touch unrelated files.

Preserve the verified Therapy filter-contract behavior, including disabled unselected zero-count topic and availability facets and the resolved focused DOM regression. For the shared docs/filter-contract.md, take current main as the base and reapply only still-current Therapy-specific rollout text; do not overwrite newer Services/Factsheets/Formulation contract changes. In shared Therapy bindings/select/search code, preserve all newer main catalogue aliases, navigation, clear-query and accessibility fixes while layering this PR’s filter semantics only.

Run the focused Therapy search filter/facet/mode-wiring tests plus the relevant accessibility selector, typecheck, and format. Report the new head and exact outcomes.

[github-actions]

Hi @BigSimmo — thanks for the @claude mention.

Open this in Claude Code on the web to continue here: https://claude.ai/code

Context: BigSimmo/Database · #1889

_{Automated backlink from the Claude backlink workflow. Claude Code can pick up the mention and push a scoped fix to the branch.}

[claude]

Claude encountered an error after 2s —— View job

---

I'll analyze this and get back to you.

[BigSimmo]

Fixed the current-head CI defects on this PR. The design-system adoption manifest/maturity snapshot now reflects that the legacy therapy-compass/filter-sheet.tsx is no longer imported, and the accessibility test now covers the truthful zero-count state: the current catalogue has no reviewed records, so Reviewed only is correctly disabled. Pushed focused updates on the current branch; no rebase or unrelated changes.

[BigSimmo]

Final review-and-fix summary

Final GitHub state

• I did not merge or close this pull request. GitHub records that PR #1889 was merged externally at 2026-08-13 03:36:42 UTC while this review was in progress.
• Final PR head: 5c165f227ddcea3135d4ed52f0d69813389fc466
• Base at merge: 2ae5a0aa5d339a7dc9089db134c2d9d0220444ae
• Immediately before merge, the branch was 15 commits ahead, 0 behind, with the latest base as its merge base and a clean merge state.
• Merge commit recorded by the raw PR endpoint and current main: ad2839aeebca6798f62d91e98ae69c8cd38e5aa2

Findings and disposition

• No P0/P1 defect was identified.
• The existing P2 review item about zero-count facets had already been addressed on the final head. Its only inline thread is now resolved and outdated.
• A distinct adversarial pass identified one additional reproducible P2 defect in src/components/therapy-compass/bindings.tsx: when a non-empty query is cleared, queryMatches immediately switches to the full catalogue while searchResults still uses the deferred prior query for one render. Facet counts can therefore transiently disagree with the visible result set, contrary to the filter-count contract.
• A minimal two-branch reorder and focused regression test were prepared and verified locally against the exact PR head before the external merge. They were not published after merge, because pushing to a closed PR branch or creating a workaround PR would have exceeded this task's authority. The defect remains present in the merged main tree and requires a separately authorised follow-up change.
• No unrelated or pre-existing debt was changed.

Adversarial review and local evidence

• CodeRabbit could not run because the repository bot reported a review spending/usage limit, and the local CodeRabbit CLI was unavailable. A separate manual adversarial pass was completed instead, covering deferred-state consistency, OR-within-group/AND-across-group semantics, disabled-option escape paths, accessibility behaviour, hidden filter state, and missing-test risk.
• Proposed-fix red/green/falsification proof: the regression failed on the original implementation, passed with the minimal fix, failed when the fix was reverted, and passed again when restored.
• Focused suite: 4 files, 26 tests passed.
• Repository PR-local planner: 6/6 checks passed.
• TypeScript, changed-file ESLint, Prettier, and git diff --check: passed.
• Independent projection cross-check: 9,216 topic projections, plus availability projections, agreed with searchTherapies across the real catalogue.

Exact-head CI

• Required aggregate PR required: success on exact head 5c165f227ddcea3135d4ed52f0d69813389fc466.
• CI workflow run 31663671475: completed successfully.
• Production UI was not selected for this head. No such exact-head check run existed.

Advisory and infrastructure state

• Advisory PR mergeability: success on the exact head.
• No exact-head Ensure auto-merge check run existed at the final snapshot.
• Supabase preview was intentionally ignored because the PR changed no supabase files.
• CodeRabbit review remained unavailable because of its external usage cap.

Threads and ledger

• Review threads: 1 total, 0 actionable, 1 resolved/outdated.
• No immutable heavy-review record was found for this exact final head. Repository policy prohibits a ledger-only tip, and the PR had already merged, so no post-merge branch or main mutation was made merely to add a record.

I did not merge the PR. Because GitHub had already merged it externally, final merge was no longer left to the user.

[BigSimmo]

Final review-and-fix status for PR #1889

• PR state: already merged before this finalisation pass. Final reviewed head: 5c165f227ddcea3135d4ed52f0d69813389fc466. Base at PR snapshot: 2ae5a0aa5d339a7dc9089db134c2d9d0220444ae. Merge commit: 9a0a00be33dedb01e9d59e81f42225cc6f9d3939.
• Merge/sync state: no further sync or branch mutation performed because the PR is closed/merged; post-merge edits would violate the requested stop condition.
• Review finding: the P2 zero-count facet issue was addressed in the final head. Unselected zero-result Topics, Review status, and Handout options are now disabled while selected options remain removable.
• Review threads: 1 inline Codex thread; resolved and outdated. No actionable unresolved review threads remain.
• Adversarial pass: re-checked the final head specifically for the reported dead-end facet behavior and verified the focused guard is present for all three relevant facet groups. No additional high-confidence PR-introduced P0/P1 defects were identified in this finalisation pass.
• Local verification: no new local test run was started after discovering the PR had already merged. The final head's exact-head CI is the decisive evidence for this stop-state review.
• Exact-head CI: CI, SAST, and Secret Scan workflow runs completed successfully. Production UI critical and Production UI (1), (2), and (3) all completed successfully. Build, Static PR checks, Unit coverage, Safety/config, PR mergeability, PR policy, Semgrep/Semgrep ingestion, Gitleaks, and GitGuardian also completed successfully.
• Advisory checks: Lighthouse budget (advisory) succeeded. Visual baselines (advisory) and Advisory UI were skipped. Other scope-inapplicable jobs such as migration replay and container images were skipped.
• Blockers/residual risk: none identified that can act on this already-merged PR. No permission workaround, rebase, force-push, deployment, release, or production test was performed.

The PR was already merged before this pass. I did not merge it or alter its merged state.