vibe-review
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from git diffs, logs, and external documentation files, creating a surface for indirect prompt injection.
- Ingestion points: The skill ingests data via git diff, git log, and file reads from docs/, specs/, or .agents/plans/ as defined in SKILL.md.
- Boundary markers: The instructions for the Standards, Spec, and Risk sub-agents do not specify robust boundary markers or 'ignore embedded instructions' wrappers for the injected diff content.
- Capability inventory: The skill utilizes git for data retrieval and the Agent tool to spawn sub-agents; it is explicitly defined as read-only.
- Sanitization: No sanitization, escaping, or filtering of the processed git content or documentation is described.
Audit Metadata