avoiding-false-positives
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Prompt Injection] (LOW): The skill contains directives such as 'DO NOT create the finding' and 'assume the developer knows something you don't'. These function as policy overrides that could lead an agent to suppress legitimate security warnings if they appear to follow existing insecure patterns in a codebase.
- [Indirect Prompt Injection] (LOW): 1. Ingestion points: Processes source code and code review findings. 2. Boundary markers: None present; the skill lacks delimiters to separate untrusted code comments from the agent's logic. 3. Capability inventory: The skill only influences decision-making logic and has no direct code execution or network capabilities. 4. Sanitization: None; the logic relies on trust and consistency with existing code, creating a surface where existing vulnerabilities can justify new ones.
Audit Metadata