browser-testing-with-devtools

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

• Third-party content exposure detected (high risk: 0.85). SKILL.md describes a workflow where the agent reads browser runtime content (DOM text, console logs, network response bodies, and JavaScript execution output) as “untrusted data,” which would be ingested into the agent’s LLM context via the Chrome DevTools MCP tools.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's installation instructs using npx to fetch and run "chrome-devtools-mcp@latest" at runtime ("args": ["-y", "chrome-devtools-mcp@latest", "--isolated"]), which causes remote code to be fetched and executed on startup (SKILL.md:29-31).