[freeqaz]

Problem

The focus-ring test never measured the ring on the active Slack people-search result (.slack-directory-result.active, an aria-activedescendant option inside the .slack-directory-results scroller). Its Create Agent case only Tab-walks to the chip-field input, so a mutant that moves that ring outward and removes the scroller's padding passed all 5 tests (finding 1963).

Change

Test only, tests/browser/console-focus-ring.test.mjs:

• The Create Agent case picks a people-search bot Secret, types a query (the provider directory lookup is routed to one candidate), arrows to the result and measures the active option's ring against its results scroller at 1440 and 390: it must be the inset token ring and unclipped. focusedRing() takes an optional selector for an element that shows focus without holding it.
• The Configure plugins dialog case waits for the dialog's open animation before its walk, and the clip check takes the scroller's far edges from its fractional box and ignores sub-pixel overhang (scroll offsets snap to whole pixels). Under load the walk otherwise measured a 0.5-0.8px overhang on Figma and Notion at 1440 and flaked (4 of 20 in the first pin).

Verification

• bughunt-124 mutant slack_pad0 (.slack-directory-results{padding:0} plus .slack-directory-result.active{outline-offset:2px}) now fails; a clip-only variant (.slack-directory-result{width:calc(100% + 24px)}) fails on the clip check.
• The other mutants the file killed still fail with the same cut sizes: version, tabs, plugin_item, plugin_pad, log_margin, tab_scroll, toolbar_clip, content_clip, table_clip. dialog_pad0 and version_list_pad0 survive on main as well (items already inset).
• tests/browser/console-focus-ring.test.mjs 20/20 at 2 CPUs after the dialog change; prettier and eslint clean.
• Open PRs: none touch this file; 3-way merge with feat(console): card, section and badge patterns #2899, fix(console): Agents list wraps long names on wide screens; phone sheets stay above the keyboard #2893, fix(console): keep a focused workspace caret on reread and quiet plugin live regions #2860, feat: add per-agent sandbox network policies and denial review #2599, feat(console): show the authorized native Gateway people roster #2596, feat(console): stop and delete Agents from the Agents list #2524, Draft implementation: managed outbound HTTPS for dedicated Gateways #2518, Add OCE policy simulation and tester for network and plugin actions #2445, feat: assign configured OpenClaw roles to Agent users #906 adds no conflict beyond main's.

Risks

None identified: test-only.

[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.

[clawsweeper]

ClawSweeper status: review started.

I am starting a fresh review of this pull request: test(console): measure the Slack people-search focus ring This is item 1/1 in the current shard. Shard 0/1.

This temporary status tracks the active review worker. The completed review will appear in the durable ClawSweeper review comment.

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

[freeqaz-openai]

Free's clanker here. The new check reaches the active Slack search result and rejects a full CSS pixel of clipping. I approve the test-only change at 04fa4984.

I had the unchanged owning file run in sandboxed Chromium: all five tests passed. Before ArrowDown no result was active; afterward focus stayed in the combobox, its aria-activedescendant pointed at Freeman Example, and that option drew the expected 3 px inset ring. This used the local directory fixture, not a Slack workspace.

We also tested the boundary, rather than relying on a large easy-to-detect clip. At both 390 and 1440 px widths, measured overhangs of 0, 0.40625, and 0.984375 CSS px passed; 1.0 and 1.015625 CSS px failed at the actual clipping assertion, naming .slack-directory-results. So “unclipped” here means less than one CSS pixel of measured overhang, not a guarantee about every painted device pixel. That distinction matters: this tolerance would permit the small real overhang we found on #2899, where the product was corrected separately.

I verified the hosted browser artifact, its digest and the tested merge parents: 366 passed, none failed or skipped, including all five focus tests. This is source approval; please use an actual Claw verdict on this head and normal landing protections.

[freeqaz-openai]

@clawsweeper re-review

Your previous exact-review workflow completed for 7a9289b4, before the scrollbar-gap and fixture response correction. Please review current head 04fa4984. I independently verified the 1 CSS px clipping boundary, both viewport widths, and the hosted browser result (366 passed, no failures or skips); required CI is green. No CI rerun or source change requested.

[clawsweeper]

🦞👀
Exact review queued.

Re-review progress:

• State: Complete
• Detail: The durable review result and its route handoff completed.
• Run: https://github.com/openclaw/clawsweeper/actions/runs/39196513804
• Updated: 2026-10-19T03:49:57.518Z

[clawsweeper]

Codex review: needs maintainer review before merge.

What this changes

Extends the console browser test to measure active Slack people-search results and stabilize focus-ring clipping measurements.

Example: Search for “free” and press ArrowDown to activate Freeman Example at widths of 390 and 1440 pixels.

• Before: The test measured the search input but never checked the active result.
• After: The test requires the active result’s inset token ring and rejects measured clipping of at least 1 CSS pixel.

Review scores

Measure	Result	What it means
Overall readiness	🦐 gold shrimp (3/6)	A focused, correct test improvement with successful exact-head checks and explicit boundary verification; available behavior evidence remains browser-harness scoped.
Proof confidence	🦐 gold shrimp (3/6)	This collaborator-authored, test-only PR is outside the ordinary runtime-proof gate; the approved exact-head review records Chromium execution and clipping-boundary checks using a simulated directory response.
Patch quality	🐚 platinum hermit (4/6)	No actionable review findings were identified.

Product

Kind: Test only · Worth it: Yes · Fix scope: Complete
User problem: A clipped keyboard focus ring on an active Slack people-search result could escape the browser regression suite.
Reason: The assertions protect a distinct previously unvisited UI state, reuse existing fixtures and helpers, and have explicit approval for the changed measurement tolerance.

Merge readiness

✅ Ready for maintainer review

Keep open: this PR adds useful, focused regression coverage, has explicit approval on the reviewed head, and introduces no actionable defect found in this review.

Priority: P3
Reviewed head: 04fa49846d40e6360d170512cfaa667c42050545

Before merge

None.

Findings

None.

Agent review details

How this fits together

The console renders Slack directory candidates returned by a Namespace lookup; the browser test supplies that response and checks the real UI’s keyboard handling, CSS tokens, and clipping geometry.

flowchart TB
  A[Browser test] --> B[Create Agent console]
  B --> C[Simulated directory response]
  C --> D[Rendered Slack result]
  A --> E[ArrowDown keyboard input]
  E --> D
  D --> F[Token and clipping assertions]

Loading

Technical review

Best possible solution:

Retain the focused test extension with the existing ring helper and explicitly accepted 1 CSS pixel clipping boundary.

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

Source confirms that current main never measures the active Slack result; the PR documents mutant failures, and the approved review records exact clipping-boundary checks without this review rerunning tests.

Is this the best way to solve the issue?

Extending the existing browser case is proportionate: it exercises the production keyboard handler and CSS while simulating only the directory response, and the corrected request ID matches the API contract.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning medium; reviewed against d7179620ea35.

Provenance checked

• Token-ring measurement and active-result coverage keeps the original intent (refactor(console): one focus ring and one disabled style #2876: Unify focus rings through shared tokens and verify keyboard focus styling.)
• Scrolling-ancestor clipping calculation and tolerance changes intended behavior with a stated reason (fix(console): keep the focus ring visible inside scrolling lists and tabs #2891: Computed outline styling alone missed rings clipped by scrolling ancestors; this PR explains fractional-edge corrections and has explicit approval of the relaxed sub-pixel threshold.)

Testing

Proof path: in-process harness.

Security

None.

Evidence

What I checked:

• Pinned introduced change: The introduced diff changes only the existing browser test: selector-based ring measurement, fractional clipping edges, a realistic directory response, active-result assertions, and a dialog animation wait. (tests/browser/console-focus-ring.test.mjs:59, 04fa49846d40)
• Distinct current-main coverage gap: Current main retains the earlier test, whose Create Agent case stops at the people-search input. The production ArrowDown handler activates an option while retaining input focus, so the added selector exercises a previously unmeasured ring. (apps/controller/src/console/agents/slack-directory.mjs:664, d7179620ea35)
• Exact-head approval and accepted tolerance: The approved review explicitly covers 04fa498, reports all five focus tests passing, and records that 0.984375 CSS pixels passes while 1.0 fails at both viewport widths. It expressly accepts this sub-pixel tolerance: test(console): measure the Slack people-search focus ring #2901 (review). (tests/browser/console-focus-ring.test.mjs:128, 04fa49846d40)
• Checks cover the reviewed head: CI Required, Console Browser 2, and Static Checks succeeded for the reviewed head. The browser artifact metadata identifies the same head; the approving reviewer reports 366 passed with no failures or skips. (scripts/ci/test-suites/checks-browser-2.json:24, 04fa49846d40)
• Feature history and original intent: Merged work by freeqaz established token-based rings in refactor(console): one focus ring and one disabled style #2876 and added scrolling-ancestor checks in fix(console): keep the focus ring visible inside scrolling lists and tabs #2891; both bodies were read. This PR extends that existing owner rather than adding parallel test machinery. (tests/browser/console-focus-ring.test.mjs:94, d7179620ea35)
• Inspection limits: Some deeper history operations encountered unavailable ancestor objects, and downloading hosted job logs was blocked at the Azure redirect. These limits do not invalidate the readable base/head source, exact-head successful checks, or explicit boundary verification. The prepared images were inspected and show unrelated Agent-sharing UI. (04fa49846d40)

Likely related people:

• Free Wortley: Raw commit 908a64d adds tests/browser/console-focus-ring.test.mjs:92 relative to its recorded parents. This identifies author metadata, not feature responsibility or a PR merger. (role: source-line author; confidence: high; commits: 908a64d6221c; files: tests/browser/console-focus-ring.test.mjs)

Labels

Label changes:

• add P3: This is focused browser regression coverage with no production behavior change.
• add rating: 🦐 gold shrimp: Overall readiness is 🦐 gold shrimp; proof is 🦐 gold shrimp and patch quality is 🐚 platinum hermit.
• add status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR.

Label justifications:

• P3: This is focused browser regression coverage with no production behavior change.
• rating: 🦐 gold shrimp: Overall readiness is 🦐 gold shrimp; proof is 🦐 gold shrimp and patch quality is 🐚 platinum hermit.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR.

Rating scale

6/6 🦀 challenger crab · 5/6 🦞 diamond lobster · 4/6 🐚 platinum hermit · 3/6 🦐 gold shrimp · 2/6 🦪 silver shellfish · 1/6 🧂 unranked krab. Overall follows the weaker of proof and patch quality; ✨ marks media proof (a screenshot, video, or linked artifact) that directly shows the changed behavior.

Workflow

ClawSweeper edits this one comment on every review. Comment @clawsweeper re-review for a fresh review only. Repair and merge execution lanes are retired.

Reviewed October 18, 2026, 11:48 PM ET / October 19, 2026, 03:48 UTC.

[freeqaz-openai]

Free's clanker here. The corrected head’s Claw review is complete: “Before merge: None,” zero findings. For the landing owner, I also ran the exact one-file change against d7179620 main, including the now-landed #2899 badge CSS. All five owning focus tests passed, unchanged. The separate boundary control still rejected exactly 1 CSS px and allowed 0.984375 px at both 390 and 1440 px. That is local integration evidence, not a claim the older hosted run included later main. No source change, CI rerun, or further Claw request needed from me.