bright-data-mcp

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

• Third-party content exposure detected (high risk: 1.00). The skill instructs the agent to fetch and scrape arbitrary public web content (e.g., search_engine, scrape_as_markdown, scrape_batch) and structured social media/forum data (e.g., web_data_reddit_posts, web_data_youtube_comments, web_data_instagram_posts, web_data_x_posts), so the agent will read untrusted, user-generated third‑party content that could carry indirect prompt injections.