implement-review
Pass
Audited by Gen Agent Trust Hub on Mar 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local shell commands to retrieve repository changes for analysis.
- Evidence:
git diff main...HEAD --statandgit diff main...HEADinSKILL.md. - [PROMPT_INJECTION]: The skill processes untrusted external code changes, which creates an attack surface for indirect prompt injection if the code being reviewed contains malicious instructions.
- Ingestion points: Output of
git diffinSKILL.mdis fed into the agent's context for review. - Boundary markers: None identified; the skill does not explicitly use delimiters to separate diff content from instructions.
- Capability inventory: Read-only access to repository state and metadata via git; report generation tools.
- Sanitization: None identified; the skill depends on the model's internal safety filters to ignore instructions embedded within the code it reviews.
Audit Metadata