remotion-mcp — AI-Powered Video Creation via MCP
remotion-mcp bridges the Model Context Protocol and the Remotion video framework, enabling AI assistants like Claude, Cursor, and Windsurf to scaffold, edit, and render React-based video compositions through conversation. It features 12 tools covering the full video production lifecycle — from project scaffolding and component generation to async video rendering with live progress polling. The server is grounded in the official Remotion documentation via a proxy tool, ensuring the AI generates accurate API calls instead of relying on stale training data. Published to npm for zero-install usage via npx.
Technical highlights:
- •
Designed and published a 12-tool MCP server enabling AI assistants to render MP4s, stills, and GIFs from natural language prompts — zero manual UI interaction required.
- •
Implemented async render job architecture with UUID-based job IDs and onProgress polling, eliminating MCP request timeouts on long video renders.
- •
Built an auto-discovery loader that dynamically scans the tools/ directory at startup — contributors add a new tool file and it registers itself with no changes to index.ts.
- •
Proxied the official @remotion/mcp documentation server as a passthrough tool, grounding the AI in live Remotion API signatures before code generation.
- •
Shipped Docker-ready deployment with Chrome pre-installed, CHROME_EXECUTABLE env var support, a /health endpoint, and a complete Railway/Fly.io deployment guide.