simplify
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection surface via code review. The skill ingests untrusted data from git diffs and possesses file-write capabilities to apply fixes.
- Ingestion points: Output of
git diff(SKILL.md, Phase 1). - Boundary markers: None present to distinguish between review instructions and the code being reviewed.
- Capability inventory: File-writing and modification instructions (SKILL.md, Phase 3).
- Sanitization: No sanitization or validation of the diff content is performed before processing.
Audit Metadata