Model Context Protocol

Process video from any AI agent

The only MCP server for video processing. Use ffpipe with Claude, Cursor, Windsurf, VS Code, and n8n.

What is the ffpipe MCP Server?

The ffpipe MCP (Model Context Protocol) server lets you process video directly from Claude, Cursor, Windsurf, and other AI agents. No external tools, no manual setup — just natural language requests that turn into optimized video processing jobs.

Unlike traditional APIs, MCP integrates video processing into your AI workflow. Ask Claude to "compress this video for TikTok" and it handles everything: downloading the video, processing it with the right presets and parameters, uploading the result, and returning the URL — all without you writing a single line of code.

Installation

Add ffpipe to your Claude Desktop or other MCP client by updating your configuration file:

claude_desktop_config.json
{
  "mcpServers": {
    "ffpipe": {
      "command": "npx",
      "args": ["-y", "ffpipe-mcp"],
      "env": {
        "FFPIPE_API_KEY": "your-api-key"
      }
    }
  }
}

Get your API key from app.ffpipe.net. Store it securely — never commit to version control.

Available MCP Tools

The ffpipe MCP server exposes these tools to your AI agent:

run_preset

Execute a named preset (convert to MP4, generate thumbnail, extract audio, etc.) on a video file.

run_command

Run custom FFmpeg commands with full control over inputs, outputs, and filters.

get_job_status

Check the status and results of a video processing job by ID.

list_presets

Get a list of all available presets and their descriptions.

Supported Clients

Use ffpipe MCP with:

Claude Desktop Native MCP support
Claude Code (CLI) Full MCP support
Cursor MCP integration
Windsurf MCP integration
VS Code + Continue via Continue plugin
n8n Workflows Direct API

Real-World Examples

You
"I have a 4K video for TikTok. Can you resize it to 1080x1920 and add captions?"
Claude (with ffpipe)
I'll help you optimize that video for TikTok. Let me:
  1. Download your video
  2. Run the resize-for-platform preset (TikTok 9:16)
  3. Burn your captions into the file
  4. Return the optimized URL
You
"Compress this for web. Target file size: 5MB max."
Claude (with ffpipe)
I'll compress your video to under 5MB. I'll calculate the optimal bitrate, apply codec settings, and target your file size requirement. Processing starts now.
You
"Extract audio from this video and make it 128k MP3."
Claude (with ffpipe)
Done! I've extracted the audio and converted it to 128k MP3. The audio file is ready at [output URL]. Total size: 12.5MB.

Frequently Asked Questions

Do I need a separate API key for MCP?

No. MCP uses your existing ffpipe API key. The same credentials work for direct API calls, n8n workflows, and MCP clients.

Which MCP clients are officially supported?

ffpipe MCP is tested and supported on Claude Desktop, Claude Code, Cursor, Windsurf, and VS Code with the Continue plugin. Any MCP-compatible client can theoretically use the server.

Is MCP included in the free tier?

Yes. The free plan includes 100 minutes of video processing per month. MCP uses the same quota as API calls and n8n workflows.

Can I run ffpipe MCP offline?

No, the MCP server requires internet connection to reach ffpipe's processing servers. Video processing happens on ffpipe's infrastructure.

How do I update the ffpipe MCP server?

MCP automatically fetches the latest version via npx. No manual updates needed. Restart your client to get the latest tools and features.

Can I use MCP with custom FFmpeg commands?

Yes, the run_command tool lets you pass custom FFmpeg arguments. Build complex workflows directly from your AI agent.

Ready to process video with AI?

Get started with 100 free minutes per month. No credit card required.

Get Free API Key

Need direct API access? See the API documentation or explore available presets.