Best possible solution:

Land the injected reader seam once exact-head CI confirms the combined agents-core shard, keeping the canonical reader as the unchanged production default and the isolated mock confined to tests.

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

Yes. The linked main workflow failure demonstrates the shared-worker import-order problem, and the grouped orphan-recovery plus registry test command is the focused reproduction path.

Is this the best way to solve the issue?

Yes. Dependency injection at the existing orphan-recovery boundary is narrower and more maintainable than preserving a module-cache mock or adding module resets, while leaving production composition unchanged.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 0471b293bb5c.

Label justifications:

• P3: This is a small, behavior-neutral test-isolation and CI reliability fix with no demonstrated user-facing runtime impact.
• rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🌊 off-meta tidepool and patch quality is 🐚 platinum hermit.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Not applicable: The protected maintainer PR changes test isolation and CI determinism without user-visible runtime behavior; focused test output and exact-head CI are the appropriate proof path.

PR surface:

Source +6, Tests +13. Total +19 across 2 files.

What I checked:

• Failure evidence: The PR identifies the failing main workflow run and a grouped two-file execution that exercises the shared-worker module-cache order responsible for the failure. (0471b293bb5c)
• Owner-boundary fix: The head adds one optional transcript-reader dependency at the orphan-recovery boundary, defaults it to the canonical production reader, and uses the selected reader only at the existing transcript-read call. (src/agents/subagent-orphan-recovery.ts:234, 198b41448597)
• Scheduling continuity: The same dependency is threaded through scheduled retry attempts, so tests can isolate every recovery attempt without changing the production default or retry lifecycle. (src/agents/subagent-orphan-recovery.ts:595, 198b41448597)
• Test isolation: The tests replace the transcript module mock with a resettable function passed through their local recovery and scheduling helpers, removing dependence on prior imports in the shared Vitest worker. (src/agents/subagent-orphan-recovery.test.ts:30, 198b41448597)
• Scoped policy alignment: The scoped agents policy discourages broad partial mocks and module resets in hot agent tests and recommends explicit dependency injection or narrow helpers. (src/agents/AGENTS.md:3, 0471b293bb5c)
• Feature provenance: Merged pull request fix: resume orphaned subagent sessions after SIGUSR1 reload #47719 introduced the orphan-recovery module and its original tests in merge commit 680eff6. (src/agents/subagent-orphan-recovery.ts:1, 680eff63fbf8)

Likely related people:

• joeykrug: Authored the merged orphan-recovery implementation and its initial test coverage in fix: resume orphaned subagent sessions after SIGUSR1 reload #47719. (role: introduced behavior; confidence: high; commits: 680eff63fbf8; files: src/agents/subagent-orphan-recovery.ts, src/agents/subagent-orphan-recovery.test.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 2026-07-17T08:49:02.628Z sha a015350 :: needs real behavior proof before merge. :: none
• reviewed 2026-07-17T08:56:37.011Z sha fd7d864 :: needs real behavior proof before merge. :: none
• reviewed 2026-07-17T09:09:34.662Z sha 9b50fbd :: needs real behavior proof before merge. :: none