fix(handoff): flag injection payloads in content-guard lint (#477)

[solomonneas]

Summary

• Extend brigade handoff lint --content-guard with line-numbered injection heuristics alongside the existing leak scan.
• Add scan_handoff_injection_heuristics() for override phrases, fake system blocks, assistant directives, and base64+decode chains; benign prompt-injection discussion downgrades to info.
• Add an injection fixture corpus and tests; document the broader --content-guard scope in CLI help and docs/security.md.

Closes #477.

Test plan

• ./scripts/verify (4200 passed, 82.62% coverage)
• pytest tests/test_handoff_injection_lint.py tests/test_untrusted.py -q
• Evil fixtures warn with line numbers; benign fixtures pass without warnings

Made with Cursor

[coderabbitai]

Important

Review skipped

Auto reviews are limited based on label configuration.

🏷️ Required labels (at least one) (1)

• coderabbit-review

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Repository: escoffier-labs/coderabbit/.coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: 7ce5639b-0d5e-4660-b2cd-17818e187fd6

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch issue/477-content-guard-injection

---

_{Comment @coderabbitai help to get the list of available commands.}