test(web): cover activity accessibility and redirect ownership

[lidge-jun]

Summary

Ordinary dependent draft layer above #553. Two test files plus two literal Manager steer-description strings change. Runtime behavior, option values, handlers and dependencies are unchanged.

• Extend Activity view coverage for literal bounded errors, paging focus, eviction, external focus and disclosure preservation. All 13 original cases remain represented; error termination adds coverage alongside the original done case.
• Exercise real steer admission/input persistence/event-bus packets through the actual Web consumer, with negative ownership/replay metadata, late A versus live B and legacy missing-mode cases.
• Keep test ports explicit: replacement dispatch and browser transport are substituted, not claimed as an actual ACP provider or SSE socket.

Source a6fd1a9c4beb38f7fcd6434f8b4f33daff2e8df1 cherry-picked unchanged. Both file hashes match the independently reviewed subset; review found no blockers.

A/C verification (07e6d4344)

72 tests /4 files passed: activity-view21, web-cursor-redirect9, web-activity-live32 and web-native-finality10. Tests used isolated homes, bounded output/process cleanup and asserted unexpected-HTTP/dispatch ledgers. Relevant production source matches the reviewed source. No full local suite or new build was needed for this test-only layer.

Known fixture limits remain: JSDOM focus is not a real browser accessibility audit; peripheral memory sidebars are deliberately unavailable; IndexedDB, Trace drawer, actual EventSource filtering, full provider/lifecycle delivery and real ACP cancellation are not certified. Hosted CI runs on the published head; queued checks are not passes. Final Electron QA remains separate.

Steer description update (1e1b1f0db)

Also carries source 612d67d20: the Manager now distinguishes same-turn input, native Cursor/Grok cancel-and-reprompt, and other runtimes' kill/resume behavior, with explicit context-preservation limits. Only the option label and explanatory text changed; values/settings/onChange logic did not. Fresh frontend typecheck and build passed after this addition. Final Electron QA must still verify the new text's rendered layout; no older capture is claimed as proof for this copy.

Dependency

Order	PR	Scope
1	#551	Runtime integration and baseline regression repairs, including FS hydration coverage
2	#552	DOM reading/focus order
3	#553	Persistence fixture drain and Windows SDK test selection
4	This PR	Activity-view and redirect ownership oracles

Merge bottom-up after per-head review/CI/impact-QA gates. No native stack, old-donor closure, auto-merge, release or deployment. Unready process-harness work is excluded.

Windows physical-fixture repair (1a0e9b5)

Carries the reviewed session-test repair from 8269b2679 in lower #553 and normal parent merges upward. Real held children replace an autonomous 2-second exit; physical factory cleanup uses the unchanged production default budget while synthetic 100ms fixtures remain unchanged. Exit/active-count assertions run immediately at the production return boundary, before an independent close wait; a foreign PONG sentinel verifies sibling preservation.

Fresh scoped verification at this head: 52 tests /2 files passed. The only change from this PR's previous head is claude-sdk-session.test.ts; production termination policy and the complete 15-file Windows selection are unchanged. Actual new-head Windows and Node CI are required and tracked separately. Earlier test/runtime receipts above are not relabeled as fresh runs on this head. No force-push, rebase, skip or gate waiver.

Rendering and bootstrap repair update (3b1d622)

Includes the reviewed Node A repair from 916653397: behavioral badge/final-render proofs, actual Mermaid callback dispatch, settings-failure/location ordering and AST-verified API counts. These repairs are present in lower #551 and propagated normally; the required existing replacement tests are retained there. Each layer freshly passed 60 tests across six named files after that update.

Also includes the reviewed Node B repair from be4f5b3b: observable reconnect/history/sidebar readiness, declared HTTP ledgers and an actually-started stale-A request. Model-free/cache/request-bridge repairs start in #551; the newer redirect fixture is repaired only in its existing #554 layer and descendants. No new test is introduced into an earlier layer merely to apply a patch.

Fresh bootstrap verification at this head: 70 tests /7 files passed. The test-only repairs preserve production code, process policy and Windows selection. Exact-head hosted Node/Windows results remain mandatory; previous Windows successes are not relabeled as new-head CI. No whole-suite local run, force-push, rebase, skip or approval bypass.

Shared disconnect repair (591f841)

This head carries the shared lowest-layer repair through normal parent merges: disconnect cleanup uses the existing synchronous UI import, and a same-stack exact-effect regression preserves every prior oracle. Fresh Node22 focused verification at this head: 80 tests across 5 files, all passed with closed processes and removed owned roots. Historical CI failures remain anchored to their original heads; fresh hosted CI, other runtime reviews and applicable QA are still required. This update is not final merge approval.

Claude cleanup and steer repair (690e6ca)

The shared repair is applied at #551 and normally propagated here. Main acquisition cleanup stays tracked after logical completion; late fulfillment finishes only its captured control and worker directory cleanup remains worker-only. Main, slash fallback and queued steer wait for main accounting without waiting for intentionally surviving workers; inclusive shutdown and exit-settle/salvage order remain intact. Fresh Node22 focused verification at this exact head: 168 tests across 13 files, all passed with observed process closure and owned-root cleanup. No SDK, protocol, permission or delivery-contract expansion. Other runtime/platform reviews, fresh hosted CI and applicable QA remain required; this is not final merge approval.

Pi preparation and cleanup repair (b76b7bc)

Carries the lowest-layer Pi repair through normal parent merges: one asynchronous completed version observation per RPC gates prompt dispatch while the actual RPC child is returned immediately. Both pool capability getters remain live. Direct Stop and RPC exit start the same bounded paired cleanup; first failure claims its result before draining, and temporary worker deletion requires both immutable physical evidence and captured directory identity. Fresh Node22 focused verification at this exact head passed223 tests across11 files with closed processes and removed owned roots. Typed finality is preserved. The older3s command-availability resolver, opaque/escaped wrapper limitations, aggregate shutdown and actual Windows/provider/native-GUI proof remain distinct limits. Fresh hosted CI and applicable review/QA are still required; no final merge approval.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Team

Run ID: 13a09461-394f-4033-896c-d8afc732244b

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-07T11:24:46.790226Z	96daa1e	Draft marked ready

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.