code-quality
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source code and pull request content for quality review and self-evaluation loops.
- Ingestion points: Source code files and PR descriptions are analyzed using the instructions in
SKILL.mdand the providedscripts/review-checklist.pytool. - Boundary markers: The skill lacks instructions to wrap reviewed content in explicit delimiters or to provide the agent with warnings to ignore embedded instructions found within comments or string literals in the source code.
- Capability inventory: The skill enables the agent to execute a local Python analysis script (
scripts/review-checklist.py) and perform iterative refinement cycles using "LLM-as-judge" patterns as described inSKILL.mdPart 3. - Sanitization: No logic is provided to sanitize, filter, or escape potential prompt injection content (e.g., instructions hidden in code comments) before they are processed by the agent's evaluation logic.
Audit Metadata