What problem does it solve? When an agent needs to answer questions requiring external knowledge β identifying a specific object, person, or place in an image or video, or confirming facts that cannot be determined from media alone β it needs grounded web evidence instead of guessing from appearance. ## Core Features & Use Cases - Web Search: Find facts using web_search with automatic backend selection across Serper, Tavily, Exa, or Serply based on configured API keys. - Page Extraction: Read a promising page in depth with web_extractor to gather detailed evidence. - Reverse Image Search: Identify entities from a frame or photo with image_search (Serper Lens), with optional bbox in 0β1000 coordinates to isolate the subject. - Use Case: While analyzing a video, grab a frame with the core capability's save_view, run image_search to identify an aircraft model, then confirm details with web_search and web_extractor before answering. ## Quick Start Set a SERPER_API_KEY (or Tavily, Exa, or Serply key) and ask the agent to verify what is shown in this video frame using web search before answering.