webperf-core-web-vitals

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

• Third-party content exposure detected (high risk: 1.00). The SKILL.md explicitly instructs the agent to navigate to an arbitrary target URL via mcp__chrome-devtools__navigate_page and then run in-page snippets with mcp__chrome-devtools__evaluate_script (e.g., LCP-Image-Entropy.js, LCP-Trail.js) that read DOM elements, resource URLs and performance entries from third-party webpages, and those extracted values drive the decision tree to run follow-up analyses — exposing the agent to untrusted, user-controlled web content that can influence actions.