cmd-pr-review-prepare
Pass
Audited by Gen Agent Trust Hub on Aug 8, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted data from git branches via
git diff. This creates a surface for indirect prompt injection where malicious instructions embedded in code comments or strings could attempt to manipulate the agent's behavior during the review process. This is a common characteristic of code analysis tools. - Ingestion points:
SKILL.mdinstructs the agent to read and build context from the output ofgit diffcommands. - Boundary markers: Absent. There are no instructions to the agent to disregard instructions found within the code being analyzed.
- Capability inventory: The agent is authorized to perform file modifications (‘Make cosmetic changes’) and suggest architectural updates.
- Sanitization: None. The skill relies on the underlying LLM's safety filters when processing the diff content.
Audit Metadata