research-new-brand
Pass
Audited by Gen Agent Trust Hub on Jul 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No security issues detected. The skill uses standard tools for research and image generation. It follows best practices by resolving paths from user inputs and documenting licensing constraints for sourced assets.
- [PROMPT_INJECTION]: The skill processes untrusted content from the web and external videos, which is a known surface for indirect prompt injection. However, the severity is reduced to SAFE as this behavior is fundamental to the skill's primary research purpose. Ingestion points: WebSearch and WebFetch in Step 3, video analysis in Step 4. Boundary markers: Not specified for prompt interpolation. Capability inventory: File system write access for brand documentation and manifests, plus usage of image generation atoms. Sanitization: No explicit sanitization of retrieved content is mentioned, but the skill's focus on structured extraction (pricing, benefits, audience) provides a functional limit to adversarial instructions.
Audit Metadata