Best possible solution:

Keep the test-local capability probe and typed symlink fixture, then require a PR-body evidence update with redacted Windows run output before merge.

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

Yes. Current main shows the two marketplace symlink tests are skipped on every Windows host via it.runIf(process.platform !== "win32"); I verified this from source rather than a Windows run.

Is this the best way to solve the issue?

Yes, for the code shape. After the second commit, the fixture uses the same platform-specific directory link type as the probe and matches existing sibling test patterns; the remaining gap is proof, not another code change.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 07bf384a8b3f.

Label justifications:

• P3: This is a low-risk test-portability cleanup with no user-visible runtime behavior change.
• rating: 🧂 unranked krab: Overall readiness is 🧂 unranked krab; proof is 🧂 unranked krab and patch quality is 🐚 platinum hermit.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: Missing: the PR body asserts a local test run but includes no redacted terminal output, copied live output, or artifact for the Windows after-fix path. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.

PR surface:

Tests +27. Total +27 across 1 file.

What I checked:

• Current main still has unconditional Windows skips: At current main, both marketplace symlink boundary tests still use it.runIf(process.platform !== "win32"), so main has not already implemented the requested test-portability change. (src/plugins/marketplace.test.ts:656, 07bf384a8b3f)
• PR head fixes the earlier fixture mismatch: The current PR head adds a platform-specific directorySymlinkType, gates the tests on canCreateDirectorySymlinks, and passes that same type into fs.symlink in the outside-symlink fixture. (src/plugins/marketplace.test.ts:165, 62eee7b86821)
• Existing sibling pattern supports the patch shape: src/infra/install-safe-path.test.ts already uses the same win32 ? "junction" : "dir" directory-symlink probe pattern before it.skipIf, so the PR follows an existing test-local approach. (src/infra/install-safe-path.test.ts:17, 07bf384a8b3f)
• Marketplace validation is the behavior under test: ensureInsideMarketplaceRoot canonicalizes existing paths with fs.realpath and returns plugin source escapes marketplace root when a candidate resolves outside the marketplace root. (src/plugins/marketplace.ts:955, 07bf384a8b3f)
• Previous ClawSweeper finding was addressed: The only diff between the previously reviewed SHA and the current PR head is the fixture update that passes directorySymlinkType into fs.symlink. (src/plugins/marketplace.test.ts:165, 62eee7b86821)
• Real behavior proof is still absent: The PR body asserts a local node scripts/run-vitest.mjs src/plugins/marketplace.test.ts run, but it includes no copied terminal output, log, or artifact and does not show the Windows after-fix path. (62eee7b86821)

Likely related people:

• Agustin Rivera: Authored the commit that added remote marketplace canonicalization and the symlink escape tests this PR changes. (role: remote marketplace path hardening author; confidence: high; commits: b1dd3ded3589; files: src/plugins/marketplace.ts, src/plugins/marketplace.test.ts)
• vincentkoc: Authored and merged the PR that introduced the marketplace registry install surface and its initial tests. (role: original marketplace feature author and merger; confidence: medium; commits: ff2e864c98cb; files: src/plugins/marketplace.ts, src/plugins/marketplace.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 2026-06-28T07:20:57.623Z sha 86678f0 :: needs real behavior proof before merge. :: [P2] Use the probed link type in the fixture