browser-testing-with-devtools

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill interacts with external, untrusted web content (DOM, console logs, and network responses) which could contain malicious prompts. The author provides detailed defensive instructions:
  • Ingestion points: Data retrieved through the Chrome DevTools MCP tools.
  • Boundary markers: Instructions to explicitly label browser data and never treat it as commands.
  • Capability inventory: Includes JavaScript execution, network monitoring, and DOM manipulation tools.
  • Sanitization: Guidelines mandate user confirmation for navigating to extracted URLs and flag instruction-like text found in page content.
  • [EXTERNAL_DOWNLOADS]: The installation section describes using npx -y to download and run the chrome-devtools-mcp package from the NPM registry.
  • [COMMAND_EXECUTION]: The skill involves executing a shell command to initialize the chrome-devtools MCP server.
  • [DYNAMIC_EXECUTION]: The skill allows for JavaScript execution within a browser page. Security constraints are provided to limit execution to read-only operations, block external network requests from within the page context, and prevent access to authentication materials like cookies or localStorage tokens.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 08:43 AM
Security Audit — agent-trust-hub — browser-testing-with-devtools