feat: recommend complete Featured lineups for both catalogs

[Patrick-Erichsen]

What Problem This Solves

The intelligence report lists shortlisted additions without proposing the complete eight-item Featured lineup or comparing it with the current selection.

User Impact

Every iteration proposes up to eight plugins and eight skills, with keeps, additions, removals, unfilled slots, supporting search/adoption evidence and freshness. Existing selections are reassessed; emerging labels use the existing New/Rising evidence. Dashboard, API, CLI and weekly digest share the same complete proposal.

Why This Change Was Made

Hydrate current membership independently of bounded search/Trending inspection and rank within each catalog using the existing evidence cohorts. Keep unknown evidence explicit, preserve query privacy, and never pad an incomplete lineup. A versioned v3 digest retains every selected item while compacting auxiliary evidence; deploy Hermit #45 before enabling its producer. Frozen v1/v2 contracts remain supported.

Recommendations remain advisory: the revised production lineup needs Patrick’s approval. This PR neither publishes Featured nor changes the homepage default.

Evidence

Before: current Featured members absent from inspected search/Trending could disappear from consideration; digest capped candidates at five. After: complete8 includes retained/add/remove partition and baseline, with explicit shortfall for an empty catalog. Registered Convex regression covers 101 unrelated Claws before current plugin membership. Actual authenticated UI at four viewport sizes shows the eight-item proposal; native report reads publish nothing, eighth grant succeeds, ninth fails and retain preserves timestamps. Comparable baseline/candidate browser artifacts are attached separately.

Native Monday-window reports passed through the real production digest builder and Hermit parser/Carbon renderer without delivery. Production +329 net lines for full-set recommendation/read/UI/wire behavior; tests/support +363, docs +18. Structured review and root owner-boundary review completed; discovered projection/unknown-evidence issues were corrected and verified.

Validation at the complete stack: static, 6,903 unit tests, types/build, package verification, and HTTP/CLI e2e all passed. Native Convex and authenticated browser checks use isolated labeled fixtures; production Featured and digest delivery remain unchanged.

Related: CLAW-724.

Review follow-up: 1dfa7198ef updates the existing weekly-digest and search-intelligence specifications for complete v3 lineups, compaction/failure semantics, legacy replay, cap transition and receiver-first deployment. Hermit #45 merged as 911c819; its successful Cloudflare build identifies production Worker version da09efd1-efc7-4275-be99-b4dab37b09d4. The live receiver accepted the deployed shared credential and returned expected GET 405 at 2026-09-16T03:21:18Z; no digest POST occurred. Cloudflare control-plane traffic percentage and D1 continuity were not re-read because its credentials are unavailable; neither the D1 binding nor receipt schema changed.

[clawsweeper]

🦞👀
ClawSweeper picked this up.

Pull request received. I will update this pull request when review starts.

ClawSweeper review complete

ClawSweeper finished reviewing this revision. The review result is being finalized.

View the workflow run.

[vercel]

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

Project	Deployment	Actions	Updated
clawhub	Ready	Preview	Sep 16, 2026 3:24am UTC

[Patrick-Erichsen]

ClawHub UI Proof

Status: passed
Mode: before-after
Scenario: claw724-complete-featured-lineup
Provider: isolated-local-convex-authenticated-chromium
Baseline: 59a90cc9eba6835cc564924f351666e7d923c369
Candidate: bd88ed5cc44d5c2242eb636b1f6dfd994cef555b
Trigger: CLAW-724

Complete plugin selection · desktop 1440×900	Complete plugin selection · desktop 1440×900

Complete plugin selection · laptop 1366×768	Complete plugin selection · laptop 1366×768

Complete plugin selection · tablet 768×1024	Complete plugin selection · tablet 768×1024

Complete plugin selection · mobile 390×844	Complete plugin selection · mobile 390×844

Empty skills · desktop 1440×900	Empty skills · desktop 1440×900

Empty skills · laptop 1366×768	Empty skills · laptop 1366×768

Empty skills · tablet 768×1024	Empty skills · tablet 768×1024

Empty skills · mobile 390×844	Empty skills · mobile 390×844

Raw proof files: https://github.com/openclaw/clawhub/tree/qa-artifacts/clawhub-ui-proof/pr-3726/claw724-bd88ed5

[clawsweeper]

Codex review: blocked before merge. Reviewed September 15, 2026, 11:18 PM ET / September 16, 2026, 03:18 UTC.

ClawSweeper review

What this changes

Proposes complete eight-item Featured selections for plugins and skills, including retained entries, additions, removals, and unfilled slots across the dashboard, API, CLI, and weekly digest.

Merge readiness

⛔ Blocked before merge - 3 items remain

This remains a useful, distinct contribution: current main provides shortlists rather than complete Featured lineups. The collaborator-authored PR has convincing browser evidence and no blocking code finding.

Priority: P2
Reviewed head: ea41d09efe1cc66bc4faa1582b0ebeb034cfda25

Review scores

Measure	Result	What it means
Overall readiness	🐚 platinum hermit (4/6)	A coherent implementation with strong visible and native evidence, a small specification correction, and a concrete rollout dependency.
Proof confidence	🦞 diamond lobster (5/6) ✨ media proof bonus	Sufficient (linked_artifact): Authenticated local Chromium screenshots demonstrate the changed recommendation UI, and the native receipt records unchanged membership after report reads. The captured candidate’s seven central production blobs match this head; the reported digest dry run did not exercise delivery or establish receiver deployment.
Patch quality	🐚 platinum hermit (4/6)	1 actionable review finding remain.

Verification

Check	Result	Evidence
Real behavior	Verified	Sufficient (linked_artifact): Authenticated local Chromium screenshots demonstrate the changed recommendation UI, and the native receipt records unchanged membership after report reads. The captured candidate’s seven central production blobs match this head; the reported digest dry run did not exercise delivery or establish receiver deployment.
Evidence reviewed	9 items	Policy and introduced scope: Read the complete root AGENTS.md and applicable ClawHub review and Convex guidance. No nested AGENTS.md files were found under the affected directories, and no maintainer-notes directory exists. Reviewed the pinned introduction delta; the checkout remained clean. Still necessary on main: The fetched main implementation returns bounded candidates without a complete membership baseline or retain/add/remove partition. The latest supplied release, v0.23.3, does not contain this recommendation module. Real browser and native evidence: Inspected all four locally prepared desktop/laptop before-and-after screenshots plus the linked report and native receipt. The candidate shows eight plugins, one retained entry, seven additions, and two explained removals. The report records an authenticated local-admin session at localhost:3539 with isolated Convex fixtures, an empty-skill shortfall, unchanged membership after report reads, and preserved Featured timestamps. Evidence: https://github.com/openclaw/clawhub/tree/qa-artifacts/clawhub-ui-proof/pr-3726/claw724-bd88ed5.
Findings	1 actionable finding	[P3] Update the durable weekly-digest specification for v3
Security	None	None.

How this fits together

ClawHub’s staff intelligence tools combine search demand, adoption snapshots, and current Featured membership into advisory recommendations. Staff review these recommendations, while Hermit receives the weekly digest for delivery.

flowchart TD
  A[Search demand] --> D[Recommendation builder]
  B[Adoption snapshots] --> D
  C[Current Featured membership] --> D
  D --> E[Eligibility and eight-item selection]
  E --> F[Staff dashboard and CLI]
  E --> G[Frozen weekly digest]
  G --> H[Hermit receiver]

Loading

Before merge

• Update the durable weekly-digest specification for v3 (P3) - This switches new persisted payloads to v3, but specs/search-weekly-digest.md:23–41 still specifies v2, five-item sections, and dropping candidates to meet the wire budget. Those instructions contradict the new complete-lineup invariant and could guide a later change back to truncating selections. Update the spec to cover v3 membership, compaction/failure behavior, frozen v1/v2 replay, and receiver-first rollout, as required by the root durable-intent policy.
• Resolve merge risk (P1) - Enabling this producer before the compatible Hermit receiver is deployed can cause weekly digests to be rejected and exhaust their retries; the receiver merge is verified, but production deployment is not.
• Complete next step (P2) - Correct the durable v3 digest specification and confirm the compatible Hermit receiver is deployed before activating this producer.

Findings

• [P3] Update the durable weekly-digest specification for v3 — convex/lib/searchEvidenceDigest.ts:240-241

Agent review details

Security

None.

Review metrics

Metric	Value	Why it matters
Introduced scope	+842/-132 across 23 files	The change spans recommendation computation, persisted wire contracts, staff presentation, and regression coverage.
Production versus validation growth	Production +329 net; tests/support +363 net; docs +18 net	Production growth is justified by complete-set hydration and shared dashboard, CLI, and digest behavior.

Merge-risk options

Maintainer options:

1. Verify receiver-first rollout (recommended)
   Record that the deployed Hermit receiver includes the merged v3 contract before enabling the ClawHub producer.
2. Hold producer activation
   Keep the producer inactive until receiver deployment can be established.

Technical review

Best possible solution:

Keep one shared advisory lineup builder, preserve frozen v1/v2 payloads, and coordinate v3 activation with verified receiver deployment and an accurate durable contract.

Do we have a high-confidence way to reproduce the issue?

Not applicable to a new recommendation capability; source comparison and captured browser runs establish the shortlist-to-complete-lineup change.

Is this the best way to solve the issue?

Yes, the shared recommendation owner is the appropriate layer; separate UI or digest ranking would create competing selections. The remaining specification correction is narrow.

Full review comments:

• [P3] Update the durable weekly-digest specification for v3 — convex/lib/searchEvidenceDigest.ts:240-241
  This switches new persisted payloads to v3, but specs/search-weekly-digest.md:23–41 still specifies v2, five-item sections, and dropping candidates to meet the wire budget. Those instructions contradict the new complete-lineup invariant and could guide a later change back to truncating selections. Update the spec to cover v3 membership, compaction/failure behavior, frozen v1/v2 replay, and receiver-first rollout, as required by the root durable-intent policy.
  Confidence: 0.99

Overall correctness: patch is correct
Overall confidence: 0.89

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning medium; reviewed against 5bf0026d8f6f.

Labels

Label changes:

• add P2: This is a bounded staff-curation improvement without an established urgent user outage.
• add merge-risk: 🚨 compatibility: The producer switches newly persisted digests to v3, requiring a compatible deployed receiver.
• add proof: sufficient: Contributor real behavior proof is sufficient. Authenticated local Chromium screenshots demonstrate the changed recommendation UI, and the native receipt records unchanged membership after report reads. The captured candidate’s seven central production blobs match this head; the reported digest dry run did not exercise delivery or establish receiver deployment.
• add rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🦞 diamond lobster and patch quality is 🐚 platinum hermit.
• add status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (linked_artifact): Authenticated local Chromium screenshots demonstrate the changed recommendation UI, and the native receipt records unchanged membership after report reads. The captured candidate’s seven central production blobs match this head; the reported digest dry run did not exercise delivery or establish receiver deployment.

Label justifications:

• P2: This is a bounded staff-curation improvement without an established urgent user outage.
• merge-risk: 🚨 compatibility: The producer switches newly persisted digests to v3, requiring a compatible deployed receiver.
• rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🦞 diamond lobster and patch quality is 🐚 platinum hermit.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (linked_artifact): Authenticated local Chromium screenshots demonstrate the changed recommendation UI, and the native receipt records unchanged membership after report reads. The captured candidate’s seven central production blobs match this head; the reported digest dry run did not exercise delivery or establish receiver deployment.
• proof: sufficient: Contributor real behavior proof is sufficient. Authenticated local Chromium screenshots demonstrate the changed recommendation UI, and the native receipt records unchanged membership after report reads. The captured candidate’s seven central production blobs match this head; the reported digest dry run did not exercise delivery or establish receiver deployment.

Evidence

Acceptance criteria:

• [P1] git diff --check.
• [P1] bun run ci:static.

What I checked:

• Policy and introduced scope: Read the complete root AGENTS.md and applicable ClawHub review and Convex guidance. No nested AGENTS.md files were found under the affected directories, and no maintainer-notes directory exists. Reviewed the pinned introduction delta; the checkout remained clean. (AGENTS.md, ea41d09efe1c)
• Still necessary on main: The fetched main implementation returns bounded candidates without a complete membership baseline or retain/add/remove partition. The latest supplied release, v0.23.3, does not contain this recommendation module. (convex/lib/featuredIntelligence.ts, 5bf0026d8f6f)
• Real browser and native evidence: Inspected all four locally prepared desktop/laptop before-and-after screenshots plus the linked report and native receipt. The candidate shows eight plugins, one retained entry, seven additions, and two explained removals. The report records an authenticated local-admin session at localhost:3539 with isolated Convex fixtures, an empty-skill shortfall, unchanged membership after report reads, and preserved Featured timestamps. Evidence: https://github.com/openclaw/clawhub/tree/qa-artifacts/clawhub-ui-proof/pr-3726/claw724-bd88ed5. (bd88ed5cc44d)
• Proof source correspondence: Compared GitHub tree blob identities from the captured candidate with the reviewed checkout. All seven checked production owners match exactly: artifact hydration, report orchestration, ranking, digest contract, digest projection, delivery orchestration, and recommendation UI. (src/routes/-management/FeaturedRecommendations.tsx, ea41d09efe1c)
• Dependency signal and persistence boundary: The builder now emits search_intelligence_weekly_v3, which is persisted in searchWeeklyDigests and sent through the existing Hermit delivery client. The PR explicitly requires the compatible Hermit receiver first. Existing v1/v2 validators remain in the persisted union, and frozen payloads bypass rebuilding. (convex/lib/searchEvidenceDigest.ts:241, ea41d09efe1c)
• Companion receiver verified: feat: deliver complete Featured lineups for both catalogs hermit#45 merged on 2026-09-16 at 03:10:56Z. Inspected its receiver source: it accepts v3, validates complete membership and query privacy, and retains v2 support. The similarly numbered ClawHub PR in supplied relatedItems is unrelated. (src/clawhubSearchIntelligence/evidence.ts, 911c819bc68f)

Likely related people:

• Patrick Erichsen: Raw commit eaeaec4 adds convex/featuredIntelligence.ts:43 relative to its recorded parents. This identifies author metadata, not feature responsibility or a PR merger. (role: source-line author; confidence: high; commits: eaeaec46705a; files: convex/featuredIntelligence.ts)

Rank-up moves

Optional improvements that raise the rating; they are not merge blockers.

• Update the durable digest specification to describe the complete-lineup v3 contract.
• Record compatible Hermit production deployment before enabling v3 production.

Rating scale

Score	Internal tier	Crab rank	Meaning
6/6	S	🦀 challenger crab	Exceptional readiness
5/6	A	🦞 diamond lobster	Very strong readiness
4/6	B	🐚 platinum hermit	Good normal PR; ordinary maintainer review
3/6	C	🦐 gold shrimp	Useful, but confidence is limited
2/6	D	🦪 silver shellfish	Proof or implementation needs work
1/6	F	🧂 unranked krab	Not merge-ready
N/A	NA	🌊 off-meta tidepool	Rating does not apply

Overall follows the weaker of proof and patch quality.
Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

Workflow

• ClawSweeper keeps one durable marker-backed review comment per issue or PR.
• Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
• A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
• PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
• Maintainers can also comment @clawsweeper review to request a fresh review only.
• Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
• Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
• Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

[clawsweeper]

ClawSweeper status: review started.

I am starting a fresh review of this pull request: feat: recommend complete Featured lineups for both catalogs This is item 1/1 in the current shard. Shard 0/1.

This placeholder means the worker is alive and reading the current context. I will edit this same comment with the actual review when the claws are done clicking.

Crustacean status: shell secured, claws on keyboard, evidence pebbles being sorted.