matt-retro
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes session history and primary evidence, which may contain untrusted data from previous interactions. This creates a surface for indirect prompt injection where malicious instructions in the session history could attempt to influence the agent's retrospective analysis.
- Ingestion points: The skill instructs the agent to "Read the available primary evidence for that session" in
SKILL.md. - Boundary markers: No specific delimiters or instructions to ignore embedded commands within the session evidence are provided.
- Capability inventory: The skill is restricted to generating suggestions and recommendations. The frontmatter explicitly sets
disable-model-invocation: true, and the body states "Implement changes only when they are part of the user's authorized task." - Sanitization: There are no explicit sanitization or validation steps for the session data being analyzed.
Audit Metadata