vp-retro
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill evaluates session history (e.g., corrections, repeated work) to suggest improvements to repository documentation and agent memory. This creates an attack surface where instructions embedded in the session history could maliciously influence the analysis. 1. Ingestion points: Current agent session history and interaction evidence. 2. Boundary markers: Absent; there are no delimiters provided to isolate session data from the analysis instructions. 3. Capability inventory: Recommends or performs writes to repository configuration files (AGENTS.md, CLAUDE.md) and modifies agent memory. 4. Sanitization: Lacks explicit mechanisms to sanitize or validate evidence derived from the session data.
Audit Metadata