commit
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the local codebase when scanning pending commits for secrets or credentials.
- Ingestion points: Pending commit content scanned in step 2 of SKILL.md.
- Boundary markers: None mentioned for the content being scanned.
- Capability inventory: Local script execution via Node.js, Git commit/push operations, and file system reads.
- Sanitization: None mentioned; the skill relies on a local policy script to determine risk levels.
- [COMMAND_EXECUTION]: The skill executes local JavaScript files using the Node.js runtime to enforce review policies and check receipts.
- Evidence: Instructions call
node .claude/scripts/lib/commit-review-policy.cjsand reference.claude/hooks/lib/review-receipt.cjs.
Audit Metadata