[abruption]

Summary

Measure gaps in macOS opener visibility using same-user fileport, shared-description and sandbox fixtures. The harness compares observations against explicit control/case/exit contracts and reports only unknown, reject or not_excluded; FWASLOCKED never becomes positive holder evidence.

Fileport API failures remain errors, hidden-sharer results describe incomplete release, and public summaries use a strict allowlist. Actor/test stderr stays private. Setup failures preserve earlier observations, and cleanup is settled before run_ok is written.

Add macOS 15 ARM/Intel research CI and keep policy proposals in HANDOFF.md. Refs #127; depends on #153. This PR is stacked on research/127-macos-prerequisites.

Validation

• npm ci, check, lint, test (263 pass / 3 Linux skips), package smoke and staged diff check passed.
• 80 visibility tests and 23 inherited-description tests passed.
• One local nominal run of the revised harness passed: measurement_ok, run_ok, no mismatches/errors/incomplete waits, cleanup removed. Raw SHA-256: 5133e330ad798a87ecf9120788e46ca884b7e6ec1a704bd8f5723814f00d049b.
• The first integrating nominal run exposed two overly strict bookkeeping checks; its failed result is retained and described in the README. The corrected run uses matching start/end source hashes.
• Independent read-only CLI review of the frozen handoff: Claude Code 2.1.302 (claude-opus-5-5 verified from response metadata) and Codex CLI 0.163.0 (isolated default; actual model unreported), both completed on input SHA-256 300b47a1e8564064b76a887090886303ffb88caacf0260f54987d61050224f6f.
• Integrating review then corrected remaining probe/case/exit validation, stderr capture and failure recording, with regression tests and a new nominal run. Those corrections are not claimed as part of the earlier source-only CLI review.

Limits

The local measurement is macOS 27 arm64/APFS, not native Codex behavior. Hosted macOS 15 results are separate CI evidence. A positive rule check does not guarantee each sandbox profile reproduces the same omission. Bootstrap removal can only be checked for names obtained from validated ready events. No product filter, support contract, signal path or npm artifact is changed. Policy choices in #127 remain open; no merge or release is requested.

[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 2026-10-18T09:03:18.177737Z	f038d2b	Draft marked ready
🔒 Security Review	✅ Completed 2026-10-18T09:04:37.967782Z	f038d2b	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.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a3afc38bb2

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 70bc859810

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".