Read Aloud Video through MCP
Read Aloud Video uses read_aloud_video and creates page-by-page narrated visuals.
Call kp_get_catalog first, then pass these arguments to kp_create_creation:
{ "type": "read_aloud_video", "prompt": "Explain MCP step by step in six pages", "language": "en", "aspect_ratio": "16:9", "scene_count": 6}Use only catalog-supported page counts. Do not send duration_seconds.
After completion, kp_get_creation may return video, PPT, or PDF artifacts. Supported pages can be updated through kp_edit_creation.
When to choose Read Aloud Video
Use Read Aloud Video when the explanation is naturally a sequence of short pages: an onboarding guide, a lesson, a child-friendly story, or a step-by-step policy. The prompt should give each page a job. State the audience, the order of ideas, and any terms that must remain exact. This format is a better fit than continuous scenes when viewers should be able to pause on each concept.
Create, poll, then inspect
Use the page count returned by kp_get_catalog; scene_count controls the page structure, while duration_seconds is not valid for this type. Store the job returned by kp_create_creation and poll it with kp_get_job_status. When it succeeds, inspect main_url, artifacts, scenes, and editable_actions through kp_get_creation before sharing the result.
If a page needs revision, inspect the supported actions first and edit that page rather than assuming every field can be changed. For a failed job, use its structured status to correct the request instead of submitting the same job repeatedly.
Continue from here
Quickstartfor setup and job pollingExplainer Videofor one continuous narrated explanationAI PPTwhen the primary result should be a presentation file