Videotok MCP Server

Connect AI agents to Videotok with OAuth, tools, resources, catalog ids, editing, and publishing.

Videotok has two integration surfaces.

Use the REST API when your product owns the workflow: your backend calls /api/v1, stores ids, polls status, and controls the UI.

Use the MCP server when an AI agent should operate Videotok for a signed-in user: choose templates, inspect workspace assets, create videos, edit timelines, save references, and publish completed media through tool calls.

Endpoint

https://www.videotok.app/api/mcp

The server is a remote MCP server over Streamable HTTP.

What MCP Is For

Use case	Use MCP when
Agent-operated creation	A user asks an assistant to create static ads, video ads, UGC ads, raw AI model outputs, slideshows, or faceless videos.
Guided template selection	The agent should inspect catalog ids, presets, brand kits, avatars, voices, and references before choosing the right workflow.
Reference inspiration	The user wants to save, rank, recommend, or reuse visual references for a template.
Video editing	The agent needs to upload media, create an editor project, add media, inspect tracks/items, or apply timeline edits.
Social publishing	The user wants to publish completed My Media assets to connected social accounts.

REST API vs MCP

Question	REST API	MCP server
Primary user	Your app, backend, automation, or SDK.	An AI agent acting for a signed-in Videotok user.
Base URL	`https://videotok.app/api/v1`	`https://www.videotok.app/api/mcp`
Auth	Workspace API key, usually `vt_live_...`.	Videotok sign-in through the MCP connector OAuth flow.
How work is requested	HTTP endpoints such as `POST /generations` and `GET /assets`.	Agent-friendly tools such as `create_generation`, `list_workspace_assets`, and `apply_editor_edits`.
Machine-readable helper	`GET /api/v1/openapi.json` and `GET /api/v1/skill.md`.	MCP resources such as `videotok://generation-catalog` and `skill://index.json`.
Social publishing	Not exposed as public REST endpoints.	`list_social_accounts` and `publish_media_to_social`.
Large media upload	`POST /api/v1/editor/media` supports multipart `file` and JSON `source_url` uploads.	`upload_editor_media` supports inline media or HTTPS `source_url`; agents should not pass local `file://` paths to the hosted server.