git-pr-review
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains instructions such as 'ignore previous instructions' which were flagged by static analysis. Investigation confirms these are defensive measures directing the agent to ignore malicious content potentially found in git logs, rather than an attack on the agent itself.
- [PROMPT_INJECTION]: The skill addresses the risk of indirect prompt injection from untrusted git data.
- Ingestion points: Commit logs and diff content retrieved through git commands (SKILL.md).
- Boundary markers: A dedicated 'Untrusted Input Rules' section instructs the agent to treat external text as inert data and ignore embedded commands.
- Capability inventory: The skill uses standard read-only git commands and lacks network access, file-write, or administrative capabilities.
- Sanitization: It explicitly directs the agent to ignore prompt-like text and prioritize technical diff data over potentially manipulated commit messages.
Audit Metadata