Sketchie turns a prompt, a changelog, or a doc into a narrated whiteboard explainer. One POST returns the finished video and the scene graph behind it, as JSON you can edit and re-render.
POST /v1/explainer with your topic and an optional length. You get a 202 back immediately, then poll until the record carries a video URL and its scene graph.
Point any MCP client at Sketchie and let your agent make the video as a tool call. No glue code, no polling loop to hand-write.
Generate from a shell script or a release step, the same way you would run any other build tool. Useful when the explainer belongs to a tag, not to a person.
A typed client over the same endpoints, so the scene graph arrives as a real type instead of a parsed blob you have to trust.
Every scene is addressable. The API hands you the scene graph alongside the rendered video, so a wrong sentence in scene three is a scene three problem, not a regenerate-everything problem. You send an edit instruction for that one scene and it re-renders in seconds.
That is the difference between a generator and a pipeline you can actually put in CI. A one-shot tool makes you pay the full cost again for a one-word fix. Here you fix the scene.
API access is included on every paid plan from Plus up. Generations draw from the same plan minutes the app uses, metered at 2x. Minutes are the only unit, so there is no second currency to reason about and no separate API price list to check.
Solo, the $9.99 entry tier, is app-only by design. A cheap tier with unmetered API access is a wholesale door, and we would rather keep the API positioning honest.
Length caps at 6 minutes and moves in 30-second steps. That is a pedagogy decision, not an engineering limit, and it is the same reason we do not do talking heads, live action, or stock footage. We draw. If you need a one-hour lecture recording, this is the wrong tool and we would rather say so here than after you have integrated.