vision
Pass
Audited by Gen Agent Trust Hub on Apr 7, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill processes external data by navigating to user-provided URLs and analyzing images. This creates an inherent surface area for indirect prompt injection, where content on a target website or embedded in an image could attempt to influence the agent's instructions. This is a common characteristic of vision-capable skills.
- Ingestion points: External web content and images via URLs processed in
SKILL.mdthrough the/visioncommand. - Boundary markers: Absent; the skill does not explicitly instruct the agent to ignore instructions found within the visual data.
- Capability inventory: Uses
mcp__playwright__browser_navigate,mcp__playwright__browser_screenshot, andmcp__4_5v_mcp__analyze_image. - Sanitization: No sanitization of the visual input or URL content is performed before analysis.
Audit Metadata