accessibility-check

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by navigating to and processing data from user-supplied external URLs.
  • Ingestion points: Untrusted data enters the agent context through the use of mcp__playwright__browser_navigate and mcp__playwright__browser_snapshot in SKILL.md.
  • Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions embedded within the processed web content.
  • Capability inventory: The skill utilizes Playwright MCP tools to interact with web browsers.
  • Sanitization: There is no evidence of validation or sanitization of the content retrieved from external sites.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 03:22 AM
Security Audit — agent-trust-hub — accessibility-check