find-bugs
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute
git diffandgh repo view. These are standard tools for identifying code changes and retrieving repository metadata, aligning with the skill's purpose of finding bugs. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted data from local code changes.
- Ingestion points: Reads
git diffoutput and individual local files (SKILL.md). - Boundary markers: Absent.
- Capability inventory: Reads local files and repository state; reports findings to user.
- Sanitization: Absent. This combination provides a surface for indirect prompt injection if the code being reviewed contains instructions intended to mislead the agent.
Audit Metadata