Video learning and developer access
Connect Video Moment Finder
Bring your indexed videos into a supported app to understand a lecture, inspect a diagram, or find a specific explanation. The connection uses your Video Moment Finder account and its available API units.
Connect your account
- Add Video Moment Finder in your app, or add a remote MCP server with URL https://api.videomomentfinder.com/mcp.
- Start Connect and sign in to the Video Moment Finder account that holds your videos. Remote MCP uses OAuth; you do not need to paste an API key into chat.
- Review the requesting app, available tools, operation costs, and your account allowance on the approval screen. Any available one-time trial is reported there after account verification; reconnecting does not grant another trial.
- Approve access, then return to your app. If your account has no API units, the approval screen explains the limit without starting any video operation.
This is a client-neutral remote MCP server with OAuth authorization-code and PKCE. Apps that support dynamic client registration can register automatically. Keep tokens and client secrets out of chat and public documents.
Tools and operation costs
Default rates are shown below. The approval screen reports the configured rates for your connection. Frame retrieval is charged per call, not per frame.
Upload video
Writeupload_video
Starts or completes a file upload for indexing. Use only a file you own or are authorized to use.
500 units per indexed video
Check video status
Readget_video_status
Reports whether a video is queued, processing, ready, or failed.
No units
List your videos
Readlist_videos
Lists videos available to the connected account.
No units
Find a moment
Readsearch_video
Finds timestamped matches in a ready video using a text query.
1 unit per query
Read a transcript
Readget_transcript
Retrieves spoken content with timestamps, optionally for a selected time range.
1 unit per call
Inspect frames
Readget_frames
Retrieves images at selected timestamps. High-resolution retrieval may fall back to stored thumbnails when the source is no longer retained.
1 unit per thumbnail call (up to 25 timestamps); 5 per high-resolution call (up to 8)
Start with one useful result
List your ready videos first, then choose one idea to explain. At the default rates, one transcript retrieval plus one thumbnail call uses 2 API units; one transcript plus one high-resolution frame call uses 6. Additional searches or frame calls add to the total.
Indexing a new video costs 500 API units at the default rate and is separate from learning with an already indexed video. Review the expected cost before an upload begins.
Account allowance
Website video credits and API units are separate balances. Connected apps use API units. If your account receives a one-time verified-account trial, its actual grant and shared unit balance are shown after sign-in. Prior free website processing can count toward that allowance.
A trial is not available on every deployment. Signing out, reconnecting, or creating a new API key does not reset an account allowance. When no units remain, metered operations stop; you can keep studying evidence already retrieved in your chat.
If a video or frame is unavailable
Only videos accessible to the connected account can be used. Wait for a queued or processing video to become ready, or choose another video if processing failed. A public link does not by itself make a video available through MCP. File upload requires an app that can transfer the bytes to a temporary upload URL; never send your account token to that URL.
Original uploaded video files are scheduled for deletion 30 days after upload; cleanup may complete later. Playback and high-resolution frames need the original, while stored transcripts and thumbnails remain available after source cleanup. Media links are temporary and may need refreshing independently of source retention. A learning answer should name evidence gaps, keep source timestamps, and avoid guessing details from an unreadable frame.
Try a learning prompt
List my ready videos, then help me understand one key idea from a lecture, with a timestamp and a frame where available.
Choose an available video, retrieve a small amount of evidence, and explain the idea with its source timestamp.
Explain this diagram at 03:43 in video <video_id>. What does each part mean?
Inspect the actual frame and nearby transcript, distinguish visible labels from interpretation, and identify anything unreadable.
Turn this indexed lecture into study notes with the main takeaways and source timestamps.
Use transcript evidence and selected frames. In clients that support MCP prompts, the user can also invoke lecture_notes.
Help me study this idea using the evidence we already retrieved.
Continue the explanation or a practice question using existing evidence when more video retrieval is unavailable.
REST API and CLI
Direct programmatic requests use vmf_ API keys instead of MCP OAuth tokens. The public REST API supports upload, status, listing, search, transcripts, and frames. The CLI wraps upload, status, and search.