Best possible solution:

Use one explicitly approved terminal-owner invariant—preferably shared with the canonical agent terminal-outcome normalization where practical—then rebase and prove abort-first, error-first, lifecycle-terminal, queued-followup, and restart-safe paths on the exact merge result.

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

Yes—the source-level reproduction deterministically starts dispatch, aborts the same run, and then rejects dispatch; the supplied after-fix real Gateway/WebSocket proof exercises the corresponding observable terminal results. The current exact merge result still needs rerunning because the branch is behind main.

Is this the best way to solve the issue?

Unclear—the bounded patch and proof are convincing, but repository policy favors canonical terminal-outcome normalization, so the Gateway owner should confirm that this local ownership layer is the narrowest maintainable boundary.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 154d53c4f6fc.

Label justifications:

• P2: This is a bounded Gateway correctness fix for contradictory terminal state with limited but real client and session impact.
• merge-risk: 🚨 message-delivery: An incorrect precedence can emit a contradictory late error terminal after an abort or duplicate the client-visible terminal projection.
• merge-risk: 🚨 session-state: The changed path controls dedupe payloads, persisted lifecycle terminalization, active-run cleanup, and restart-safe session admission.
• 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 (live_output): The PR supplies after-fix live output from a real local Gateway and WebSocket RPC client for all three relevant terminal orderings; the later head change is test typing only and does not alter runtime behavior.
• proof: sufficient: Contributor real behavior proof is sufficient. The PR supplies after-fix live output from a real local Gateway and WebSocket RPC client for all three relevant terminal orderings; the later head change is test typing only and does not alter runtime behavior.

PR surface:

Source +92, Tests +1402. Total +1494 across 5 files.

What I checked:

• Terminal ownership implementation: The dispatch-error path now snapshots and resolves the owner of the terminal result before asynchronous restart-safe cleanup, distinguishing explicit abort, lifecycle terminal, queued follow-up, and ordinary dispatch-error paths. (src/gateway/server-methods/chat-send-dispatch-errors.ts:59, ea1400556648)
• Pre-dispatch attachment abort: Attachment setup failure checks the active run signal and abort marker before converting the setup failure into an error, preserving the already-terminal abort path. (src/gateway/server-methods/chat-send-attachments.ts:308, ea1400556648)
• Admission cleanup ownership: The terminal-abort admission path no longer force-cleans ownership that may be retained for pending terminal projection and transcript work. (src/gateway/server-methods/chat-send-admission.ts:356, ea1400556648)
• Real Gateway proof: The supplied production Gateway and WebSocket RPC proof observed three after-fix outcomes: abort-first remained aborted, error-first remained error, and a signal-only lifecycle interruption emitted only the lifecycle terminal. The only later head commit removes a redundant test type argument rather than changing runtime behavior. (src/gateway/server.chat.gateway-server-chat-b.test.ts:1401, adc576fd19e1)
• Adjacent merged dispatch lifecycle fix: The behavior overlaps the merged dispatch-failure session-lifecycle work from Fix WebChat dispatch failure session status #84352, but that change terminalized ordinary dispatch failure and did not settle the later-rejection-after-abort ordering addressed here. (src/gateway/server-methods/chat-send-dispatch-errors.ts:59, 92c10d4edcad)
• Adjacent merged abort-marker ordering: The merged fix(gateway): ignore stale abort markers for fresh chat events #91013 established structured freshness ordering for abort markers, making its implementation history relevant to this PR's abort-owner checks without superseding this distinct rejection race. (src/gateway/server-chat-state.ts, cd98f195a709)

Likely related people:

• jesse-merhi: Their merged work in Fix WebChat dispatch failure session status #84352 established the session terminalization and sessions.changed behavior that the current dispatch-error owner logic must preserve. (role: introduced adjacent dispatch-failure lifecycle behavior; confidence: high; commits: 92c10d4edcad; files: src/gateway/server-methods/chat.ts, src/gateway/server-methods/chat-send-dispatch-errors.ts, src/gateway/server.chat.gateway-server-chat-b.test.ts)
• nxmxbbd: Beyond authoring this proposal, they previously contributed the merged structured abort-marker ordering in fix(gateway): ignore stale abort markers for fresh chat events #91013, which is directly adjacent to the terminal-owner invariant. (role: recent abort-marker and Gateway lifecycle contributor; confidence: high; commits: cd98f195a709; files: src/gateway/server-chat-state.ts, src/gateway/server-chat.ts, src/gateway/server-close.ts)

• 🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
• 🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
• 🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
• 🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
• 🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
• 🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
• 🌊 off-meta tidepool: rating does not apply to this item.

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.

• 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.

• reviewed 2026-07-05T07:40:02.876Z sha 7e05517 :: needs real behavior proof before merge. :: none
• reviewed 2026-07-05T08:01:53.389Z sha 7e05517 :: needs real behavior proof before merge. :: [P2] Fix the broadcast mock typing in the new Gateway test
• reviewed 2026-07-05T08:07:52.017Z sha 5e2378a :: needs real behavior proof before merge. :: none
• reviewed 2026-07-05T08:14:25.808Z sha 5e2378a :: needs real behavior proof before merge. :: none
• reviewed 2026-07-05T09:56:05.552Z sha 5e2378a :: needs maintainer review before merge. :: none
• reviewed 2026-07-05T10:09:11.008Z sha 5e2378a :: needs maintainer review before merge. :: none
• reviewed 2026-07-05T11:48:23.263Z sha 5c2e32c :: needs maintainer review before merge. :: none
• reviewed 2026-07-10T20:11:29.301Z sha c0d19fe :: needs maintainer review before merge. :: none