Summary
Start and poll, because the render is measured in minutes rather than seconds.
There is no interactive editing: you describe the problem, the service produces an explanation video, and query_video_status is how you find out when it is ready. That shape suits an agent workflow well — fire the request, carry on, come back for the result rather than blocking on it.
What it is
An MCP server for Scenext, a service that generates educational explanation videos from a topic or a problem. Generation is started with one call and checked with another.
What you get
- gen_video — generate a teaching video that explains a given problem or topic
- query_video_status — check on a generation already in progress
- Rendering takes a while and varies with how hard the material is, so the pair is meant to be used as start-then-poll
Requirements
A Scenext API key. Connect either locally through Python, or straight to the hosted endpoint with the key in an authorization header — some clients need that header set by hand rather than through configuration.
Setup effort
One command plus a key — uvx scenext-mcp, then supply credentials