correct
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze recent commits, reverts, review comments, and agent instruction files from the repository. This content is potentially untrusted and could be used to deliver instructions that influence the agent's subsequent repository modifications.
- Ingestion points: Reads repository metadata including commits, reverts, review comments, and instruction files (SKILL.md).
- Boundary markers: The instructions do not define boundary markers or explicit prompts to ignore embedded instructions within the analyzed repository data.
- Capability inventory: The skill grants the agent the ability to modify repository architecture, add linting/CI checks, update tests, and modify agent instruction files.
- Sanitization: There is no mention of sanitizing or validating the content of the commit messages or comments before they are processed by the agent.
Audit Metadata