media-fetcher

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

• Third-party content exposure detected (high risk: 0.90). The skill autonomously navigates and scrapes public websites and social media (see Playwright MCP examples in SKILL.md and scripts/capture_screenshot.py for browser_navigate/browser_snapshot, scripts/download_video.py for yt-dlp downloads from YouTube/TikTok/Instagram/X, and scripts/fetch_gifs.py for GIPHY API), and the agent is explicitly expected to read/interpret those third‑party pages and pick/act on results, which could allow untrusted web content to influence actions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires and uses Playwright MCP (installed/run via "npx @playwright/mcp@latest") to browser_navigate to arbitrary live webpages (e.g., browser_navigate("https://vercel.com")) and injects page snapshots/accessibility trees into the agent's runtime context so externally-fetched page content can directly control the agent's prompts/actions.