correct
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted repository metadata which could potentially contain malicious instructions intended to influence the agent's refactoring actions.
- Ingestion points: The agent is instructed in
SKILL.mdto read "recent commits, reverts, review comments, agent instruction files, agent memory of past corrections, the rule table, and comments that explain workarounds." - Boundary markers: There are no explicit instructions or delimiters defined to help the agent distinguish between data to be analyzed and instructions to be followed within these external sources.
- Capability inventory: The skill grants the agent permission to perform significant repository modifications, including changes to architecture, types, lints, CI checks, and tests as described in
SKILL.md. This implies extensive file-write capabilities and the ability to execute shell commands to run tests and linters. - Sanitization: The skill does not define any sanitization, validation, or verification processes for the metadata it ingests.
Audit Metadata