implement
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to fetch ticket content from an external issue tracker, which introduces a risk where malicious instructions embedded in the ticket data could influence agent behavior.
- Ingestion points: External issue tracker data fetched based on ticket references (referenced in SKILL.md).
- Boundary markers: There are no explicit delimiters or instructions provided to the agent to differentiate between system instructions and data fetched from the external source.
- Capability inventory: The skill grants the agent the ability to execute code (tests and typechecking) and modify the repository state (committing work) based on the processed data (SKILL.md).
- Sanitization: No sanitization, filtering, or validation steps for the external ticket content are defined in the skill logic.
Audit Metadata