chrome-devtools

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The SKILL.md explicitly instructs the agent to read and inspect untrusted browser content — e.g., "DOM nodes, console logs, network responses, JavaScript execution results" (see "Available Tools" and "Security Boundaries") — meaning it consumes arbitrary third-party web pages and their data as part of its debugging workflow, which could embed instruction-like payloads.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's runtime configuration invokes "npx chrome-devtools-mcp@latest --isolated ..." which fetches and executes remote package code (a required dependency for the MCP server), so it brings remote-executed code into the agent runtime.