feat(agent-core-v2): approve unanalyzable bash commands in yolo permission mode

[sailist]

Related Issue

N/A — internal change; see Problem below.

Problem

In Ask When Needed (yolo) permission mode, bash commands that cannot be statically analyzed — e.g. commands containing $VAR, unterminated quotes, or deeply nested shells — still trigger an approval prompt. This interrupts batch flows that the mode is designed to run automatically: users who choose Ask When Needed have already accepted auto-approval of routine commands, and Never Ask mode already skips the guard entirely.

What changed

• The dangerous-command guard now asks in Ask When Needed mode only for commands positively identified as dangerous (shutdown, rm -rf, dd writing to devices, ...). Commands that cannot be statically analyzed fall through to the mode's auto-approval.
• Always Ask (manual) mode behavior is unchanged: unanalyzable commands still ask, and everything else reaches the fallback ask policy as before. Never Ask mode is untouched.
• Updated the existing yolo-mode unanalyzable test cases in place (no net test-count change); full agent-core-v2 suite passes (6728 tests); added a changeset.

Checklist

• I have read the CONTRIBUTING document.
• I have linked a related issue (external PRs: the issue must have a maintainer's /approve).
• I have added tests that prove my feature works.
• Ran gen-changesets skill, or this PR needs no changeset.
• Ran gen-docs skill, or this PR needs no doc update.

[changeset-bot]

🦋 Changeset detected

Latest commit: 712cb8b

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
@moonshot-ai/kimi-code	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[pkg-pr-new]

pnpm dlx https://pkg.pr.new/@moonshot-ai/kimi-code@712cb8b

npx https://pkg.pr.new/@moonshot-ai/kimi-code@712cb8b

commit: 712cb8b