Best possible solution:

Land the shared retry wrapper after maintainer acceptance of the exact-head CI evidence, or require a focused simulated 124/137 timeout proof if deterministic retry-branch coverage is mandatory.

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

No deterministic current-main reproducer was established. The linked GitHub Actions run failed in checks-windows-node-test on attempt 1 and passed on attempt 2, while source inspection shows current main exits after one timeout-style platform fetch failure.

Is this the best way to solve the issue?

Yes, this is the best scoped fix I found: it preserves the non-GNU manual watchdog needed by platform jobs and mirrors an existing sibling retry pattern. Duplicating four separate checkout scripts or switching platform jobs to GNU timeout would be less maintainable or less portable.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 29f787f10ed4.

Label justifications:

• P2: The PR addresses a real CI flake class with limited product blast radius but meaningful maintainer and contributor impact.
• merge-risk: 🚨 automation: The diff changes GitHub Actions checkout control flow where a mistake could block platform CI jobs before validation starts.
• 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 external-contributor proof gate does not apply to this maintainer-labeled MEMBER PR; PR-body validation and exact-head CI provide the normal review evidence.

PR surface:

Tests +11, Config -77. Total -66 across 2 files.

What I checked:

• Root repository policy read: Root AGENTS.md was read fully; its ClawSweeper review and CI-workflow guidance affects this review because the PR changes GitHub Actions checkout control flow. (AGENTS.md:1, 29f787f10ed4)
• Current main behavior: Current main's platform checkout helper has a 90-second watchdog and calls the fetch helper once, so timeout-style failure exits before platform tests run. (.github/workflows/ci.yml:1898, d43366b4e392)
• PR implementation: PR head keeps the manual watchdog as fetch_checkout_ref_once and wraps it in a three-attempt retry loop for statuses 124 and 137. (.github/workflows/ci.yml:1919, add5d50d9e6e)
• Alias consumers: PR head uses the anchored checkout step for macOS Node, macOS Swift, and iOS after defining it in the Windows job. (.github/workflows/ci.yml:2025, add5d50d9e6e)
• Guard coverage: The workflow guard now asserts one platform checkout anchor, three aliases, the retry loop, timeout filtering, and warning text across all four platform checkout jobs. (test/scripts/ci-workflow-guards.test.ts:398, add5d50d9e6e)
• Sibling retry pattern: Current main already uses the same three-attempt timeout-status retry shape in another manual checkout path, making this a consistent CI pattern rather than a new mechanism. (.github/workflows/ci.yml:135, 29f787f10ed4)

Likely related people:

• RomneyDa: Prior merged PRs bounded manual fetches and adjusted the platform checkout timeout in the same workflow/test surface. (role: same-surface prior fix contributor; confidence: high; commits: ed36f423dabf, 784fbcfd16e2; files: .github/workflows/ci.yml, test/scripts/ci-workflow-guards.test.ts)
• Peter Steinberger: Commit 0ee5baf added an existing custom-checkout retry pattern that this PR mirrors for platform checkout watchdogs. (role: adjacent checkout retry contributor; confidence: medium; commits: 0ee5baf6c502; files: .github/workflows/ci.yml)

• 🦀 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.