The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external websites via navigation and performance trace analysis. Ingestion points: External target URLs (Phase 1). Boundary markers: Absent. Capability inventory: Page navigation, network request listing, trace analysis, and file system reading (Phase 5). Sanitization: Absent.\n- [EXTERNAL_DOWNLOADS]: The skill references and suggests installation of the chrome-devtools-mcp package from the official ChromeDevTools GitHub repository, which is an established and well-known source.\n- [COMMAND_EXECUTION]: The skill provides configuration examples for executing the Chrome DevTools MCP server using npx, including the -y flag to bypass interactive installation prompts.

web-perf