Creatify MCP Server for AI Video Generation

Overview

The Creatify MCP Server is a lightweight implementation of a Model Context Protocol (MCP) server designed to expose Creatify-based AI video capabilities to developer workflows. It provides a small HTTP service that maps MCP-style tool calls to Creatify features such as avatar-driven video generation, URL-to-video conversion, text-to-speech (TTS), and a set of AI-assisted video editing operations. Running this server locally or in your infrastructure gives you a consistent API surface for orchestrating video generation tasks from applications, prototypes, or pipelines.

By packaging common media operations behind MCP-compatible endpoints, the server makes it easier to integrate avatar videos, automated narration, and simple edits into product flows (for example: content creation tools, marketing automation, or programmatic video testing). The server is intended for developers who want a self-hosted bridge between MCP orchestration and Creatify’s media features.

Features

• MCP-compatible server for tool-driven AI media operations
• Avatar-based video generation using Creatify avatars and scripts
• URL-to-video: convert web pages or content into short videos
• Text-to-speech with selectable voices for narration
• AI editing tools: trim, merge, replace, and revoice clips
• Local hosting and extensible tool registry for custom workflows
• Simple HTTP API and example manifest for quick onboarding

Installation / Configuration

Clone and install dependencies, then configure environment variables:

git clone https://github.com/TSavo/creatify-mcp.git
cd creatify-mcp
npm install

Create a .env file in the project root (example variables shown below). Exact names may vary—check the repository manifest or README for any naming differences.

# .env example
PORT=3000
CREATIFY_API_KEY=sk_your_creatify_key
STORAGE_PATH=./tmp/uploads
LOG_LEVEL=info

Common environment variables

Start the server in development or production:

# development with auto-reload (if configured)
npm run dev

# production
npm start

If running behind a reverse proxy or public tunnel (ngrok), expose the configured PORT and ensure callback URLs registered with Creatify match the public address.

Available Tools / Resources

The server exposes a set of tools you can call via the MCP protocol. Typical tools include:

• avatar.generate — produce an avatar-speaking video from a script or prompt
• url_to_video.generate — create a video from a target URL, using extracted content and layouts
• tts.synthesize — produce speech audio from text using Creatify voices
• edit.trim — trim a video clip to a start/end window
• edit.merge — concatenate multiple clips into a single video
• edit.replace_audio — swap audio tracks or revoice existing clips

Example manifest (mcp_manifest.json):

{
  "name": "creatify-mcp",
  "tools": [
    { "name": "avatar.generate", "description": "Avatar-based video generation" },
    { "name": "url_to_video.generate", "description": "URL-to-video conversion" },
    { "name": "tts.synthesize", "description": "Text-to-speech synthesis" },
    { "name": "edit.trim", "description": "Trim video clips" }
  ]
}

Typical request structure for an MCP call:

POST /mcp/run
Content-Type: application/json

{
  "tool": "avatar.generate",
  "input": {
    "avatar": "emma",
    "script": "Hello, welcome to our new product demo!",
    "format": "mp4",
    "resolution": "720p"
  }
}

Responses usually return a JSON object with job status and an asset URL when generation completes.

Use Cases

• Marketing avatar videos: programmatically create short, localized demo videos with an avatar reading marketing copy and a generated background track.
• Content repurposing: convert blog posts or documentation (URL-to-video) into short video summaries for social platforms.
• Accessibility and narration: synthesize voiceovers (TTS) and merge them with existing footage to add narrated versions of tutorials.
• Automated editing pipelines: trim, merge, or revoice clips as part of CI/CD for media-heavy applications (e.g., nightly builds that generate promo clips).

Tips and Next Steps

• Start by running the server locally and testing calls with curl or Postman to familiarize yourself with the manifest and tool inputs.
• Store larger assets in an external object store (S3 or similar) and keep only references in the MCP payloads to avoid filling local disk.
• If you plan to run at scale, add authentication in front of the MCP endpoints and monitor job status and queueing to handle long-running jobs.

Repository and issues: https://github.com/TSavo/creatify-mcp

For detailed examples, payload schemas, and advanced configuration, consult the repository README and the Creatify API documentation.