regulars
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with external web content, which is a potential vector for instructions embedded in third-party websites.
- Ingestion points: Browser MCP tools fetch data from a user-specified [URL] (found in SKILL.md).
- Boundary markers: No specific delimiters are used to separate webpage content from instructions, but the agent is explicitly restricted from reading source code.
- Capability inventory: The skill can navigate, click elements, take screenshots, and provide text-based reports.
- Sanitization: Content from the external site is not sanitized or filtered for instructions.
Audit Metadata