browser-testing-with-devtools
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill includes explicit security rules directing the agent to disregard any instructions or commands encountered within browser content (DOM, console, or network traffic), treating them strictly as data. This includes examples of patterns to ignore, such as 'Ignore previous instructions', which helps protect against indirect prompt injection.
- [EXTERNAL_DOWNLOADS]: The documentation describes configuring the
@anthropic/chrome-devtools-mcppackage from a trusted organization's repository via npm. - [COMMAND_EXECUTION]: The skill enables JavaScript execution within the browser context for debugging purposes, but establishes strict constraints: execution is intended to be read-only, cannot access credentials (cookies/tokens), and cannot perform external network requests from the page.
Audit Metadata