ab-testing
Pass
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent to read local context files like
.agents/product-marketing.md, creating a surface for indirect prompt injection. - Ingestion points:
SKILL.mdspecifies reading external context from local files like.agents/product-marketing.md. - Boundary markers: No specific delimiters or safety instructions are defined to isolate untrusted content from these files.
- Capability inventory: No high-risk capabilities (e.g., shell access, network exfiltration, or file system writing) were identified in any of the skill files.
- Sanitization: The skill does not implement validation or sanitization for the content read from these local context files.
Audit Metadata