For composing AI videos programmatically, the strongest fits are Replicate (which hosts video generation models like Prunaai P Video with a real API) and Rendobar (cloud video composition and FFmpeg through one API and MCP). Soundside 1 offers server-side video composition through an MCP endpoint, and Framelane 5 offers JSON-based video editing and rendering for agents over API or MCP.

1. 1

   Replicatereplicate.comverified and signed

   Hosts Prunaai P Video models that generate video from text, images or audio in a single run, and edit or replace people in existing videos via API 313.

   WebsiteDocsPricingGet started

   for agents: prunaai/p-video-edit · prunaai/p-video · prunaai/p-video-replace

2. 2

   Rendobarrendobar.comfound in code, not yet checked

   Composes video from one JSON request with clips, titles, captions and music timed automatically, and exposes job types and status checks over MCP 6.

   WebsiteDocsPricingGet started

   official:MCP server · Rendobarpackage · Rendobar

   for agents: list_job_types · get_job · get_account

3. 3

   Soundsidesoundside.aiverified and signed

   Provides a compose_video action that enriches a plan, generates assets in parallel and assembles them with transitions, reachable via an MCP endpoint 111.

   WebsiteDocsPricingGet started

   for agents: Server-Side Video Composition · Video Editing · Video Editing

4. 4

   FrameLaneframelane.iowebsite read, tools not yet checked

   Lets an agent edit a video project in JSON, preview it, and render the final video through its API or MCP 5.

   WebsiteDocsPricingGet started

   official:MCP server · FrameLane

5. 5

   Trezatrezalabs.comwebsite read, tools not yet checked

   Assembles AI video pipelines from models such as Veo and Seedance, with composition, captions and publishing, runnable on a schedule or as an API 4.

   WebsiteDocsPricingGet started