anyVideo

By Widget

Overview

Watch, tag and work with video inside ChatGPT. AnyVideo can bring up a video you're talking about, or search for them. e.g. - "@anyVideo show me a video of kittens" - "I want to watch @anyVideo about scuba diving" - "Recommend a video about AI then show me with @anyVideo" Use the tagging feature to mark key moments in the video as it plays

Tools

display-video

ChatGPT

Use to render a specific YouTube video inside the conversation when you already know the exact YouTube video ID. This is the final presentation step after the assistant has either received a concrete ID or selected one through a prior search. The rendered widget supports ordinary playback and a user-driven performance-observation workflow: editable coaching tags capture exact playback timestamps, optional notes add context, and the resulting log can be copied or explicitly sent to ChatGPT for analysis. Use it when the user wants to watch, preview, discuss, annotate, coach, or review footage in context. The app does not automatically inspect, transcribe, or understand the video, and it does not save annotations beyond the current widget. Do not use this tool as a search mechanism or a way to browse multiple candidates; call search-videos first when the exact ID is unknown. If the ID is malformed or unavailable, the tool renders the existing inline error state. Example user prompts: 1. Show the YouTube video with ID dQw4w9WgXcQ. 2. Play the video we just found here in the conversation. 3. Open that selected YouTube clip inline. 4. Let me watch this specific video without leaving ChatGPT. 5. Display the video whose ID is M7lc1UVf-VE. 6. Open this performance clip so I can tag coaching observations.

search-videos

ChatGPT

Use to search YouTube for candidate footage when the user wants to watch, annotate, coach, or review a video but no exact YouTube video ID is known. This tool returns normalized candidate metadata with IDs, titles, descriptions, thumbnails, channel names, publish dates, and watch and embed URLs so the assistant can select an appropriate result. Use it before display-video when the request is ambiguous or multiple matches need inspection. Do not use it when the user already supplied a concrete video ID, explicitly wants text only, or a plain-language response is sufficient. After searching, select the best result for the stated purpose and call display-video when the user wants the playable observation widget shown in chat. Example user prompts: 1. Find me a beginner video about making sourdough starter. 2. Look for the official video for Never Gonna Give You Up. 3. Recommend a funny kitten video for me to watch. 4. Search YouTube for a guided ten-minute meditation. 5. Find a clear explanation of how solar eclipses work. 6. Find a slow-motion tennis serve video for coaching review.