概览
What is VidMCP?
VidMCP is an MCP server that acts as a smart routing layer between AI assistants and video generation tools. It automatically selects the best model (Kling, Runway, or Fal.ai) based on the prompt, runs a full pipeline, and delivers the video without manual API setup. It is for users who want to generate videos via natural language from any MCP-compatible client.
How to use VidMCP?
Install Python 3.11+ and uv, clone the repository, run uv sync, and create a .env file with your API keys. Configure the server in your MCP client (e.g., Claude Desktop, Cursor, Windsurf) by adding the JSON config pointing to server.py. Then simply describe what you want in natural language — VidMCP automatically picks the best tools and executes the pipeline.
Key features of VidMCP
- Smart routing across Kling, Runway, and Fal.ai models
- Full pipeline: generation, transcription, audio, merge, quality check
- Memory of user style and preferences across sessions
- Raw video processing: upload, auto-transcribe, filler word detection
- BYOK model: bring your own API keys, no markup on generation
- Works with Claude Desktop, ChatGPT, Cursor, Windsurf
- 14 tools including route, generate, pipeline, process, animate, restyle
Use cases of VidMCP
- Generate cinematic videos from text descriptions
- Clean up recordings by removing filler words and transcribing
- Animate static product photos into short videos
- Apply style transfer or restyle existing footage
- Create voiceover audio and merge with generated or uploaded footage
FAQ from VidMCP
What does BYOK mean and how does pricing work?
BYOK means you connect your own API keys from providers like Kling, Runway, Fal.ai, and ElevenLabs. You pay those providers directly at their standard rates. VidMCP charges only for the intelligence layer — smart routing, pipeline management, and memory — not for video generation itself.
What happens if I don't have any API keys?
VidMCP auto-detects available keys and falls back to free options automatically. For video it uses Pollinations (free, no signup), for audio it uses Edge TTS (free). So you can start generating without any keys.
Which models are supported and what are their requirements?
Supported models include Kling 3.0 (cinematic, requires KLING_API_KEY), Runway Gen-4 (style transfer, requires RUNWAY_API_KEY), Wan 2.2 via Fal.ai (general video, requires FAL_API_KEY), and Pollinations (free, no key). Priority order for video is Kling →
