[QiaolongLi1201]

O148 (P2): deep PowerShell nesting hid writes from the scanner

The pwsh write-target scanner documented its own gap: one-level nested
execution was analyzed, deeper nesting was silently ignored — iex "& { Set-Content <moss config> x }" and & { & { & { & { … } } } } produced no
resolved target, no unresolved marker and no gate hit.

Fix (src/safety/pwsh-write-targets.ts)

• Nested bodies (& { … } / . { … } / iex literals / pwsh -c "…") are
  scanned recursively up to 3 levels (depth-capped, bounded).
• Deeper than the cap is reported unresolved (fail-closed) instead of dropped.

Tests — test/o148-pwsh-deep-nesting.spec.mjs (7)

Depth-1 baseline, depth 2 and 3 resolve the real config target; depth-2
through iex resolves; deeper-than-cap produces a gate hit (resolved target
or unresolved marker — never silence); deeply nested benign body fails
closed; one-level reads stay clean.

Revert-red (local, macOS)

Restoring the one-level guard → 7 red (incl. the iex chain and both deep
cases); with the fix 7/7. Related: aa-o130-pwsh-write-targets 153/153,
aa-o130-dollar-control, o170 39/39. npm run check green.

Branch fix/o148-pwsh-deep-nesting, base 6b40504f.

[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 2026-10-19T03:15:31.597071Z	f7674ed	PR opened

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: f7674edaf3

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]

Bound nested-command traversal as well as recursion depth

For deeply nested scriptblocks, extractNestedCommands() returns every descendant block, not just immediate children, so each recursive call scans descendants again and the depth cap does not bound the total work or accumulated results. After building, scanning '& { '.repeat(80) + 'Write-Host hi' + ' }'.repeat(80) took about eight seconds and produced 1,581,580 unresolved entries; 100 wrappers (only 613 characters) threw RangeError: Maximum call stack size exceeded at the spread into unresolved. Since findShellProtectedWrites() invokes this scanner synchronously, these inputs stall or abort protected-path checking instead of returning the intended fail-closed marker. Traverse only immediate nested bodies or enforce a shared work/result budget that stops scanning once exhausted.

Useful? React with 👍 / 👎.