VidContext + Cursor

Give Cursor's AI the ability to watch and understand video. Install the VidContext MCP server and Cursor can analyze demo recordings, UI walkthroughs, training videos, and competitor product demos — directly from your development environment.

Get your API keyView on PyPI

Setup in 3 steps

  1. 1

    Install the MCP server

    Open your terminal and run pip install vidcontext-mcp. This installs the VidContext MCP server globally so Cursor can find it.

  2. 2

    Configure Cursor

    Create a .cursor/mcp.json file in your project root with the VidContext server entry. See the configuration example below.

  3. 3

    Start using it

    Open Cursor (or restart if it was running). Ask the AI to analyze a video file. It will call the VidContext tool, process the video, and return structured results you can use.

Cursor MCP configuration

Create .cursor/mcp.json in your project root with this content:

{
  "mcpServers": {
    "vidcontext": {
      "command": "vidcontext-mcp",
      "args": ["--api-key", "vc_your_key_here"]
    }
  }
}

Replace the key with your own from the VidContext dashboard. Restart Cursor after saving.

Example prompts

Analyze the video at ./recordings/demo.mp4 and summarize what it shows
Extract the transcript from ./assets/tutorial.mov and add it to the docs
Watch ./competitor/product-launch.mp4 using competitor mode and list their key claims
Analyze the UGC video at ./uploads/review.mp4 and tell me if it meets our quality bar

Developer use cases

Documentation from demos

Record a product demo, then ask Cursor to analyze it and generate docs. Transcripts and visual descriptions become the basis for user guides or changelogs.

Bug report triage

When users submit screen recordings of bugs, Cursor analyzes the video and extracts what happened, what was on screen, and what errors appeared.

Training content to knowledge base

Analyze training or onboarding videos to extract key information, then build searchable knowledge base articles from the results.

Competitive intelligence

Analyze competitor product videos in competitor mode. Get structured breakdowns of features, messaging, and positioning for roadmap planning.

What VidContext returns

  • Metadata — duration, resolution, format
  • Transcript — full timestamped speech
  • Visual scenes — what appears on screen, frame by frame
  • On-screen text — titles, captions, URLs, error messages
  • Brands detected — logos and product mentions
  • Audio analysis — music, effects, and ambient sound

Frequently asked questions

How do I set up MCP in Cursor?

Create a .cursor/mcp.json file in your project root with the VidContext server entry and your API key. Cursor detects it automatically on next launch.

What can I ask Cursor to do with video?

Analyze a video and return a transcript, visual descriptions, on-screen text, brands, or use any of the 8 analysis modes. Results come back as text Cursor can reference in code or docs.

Does it slow down my IDE?

No. The MCP server only runs when Cursor's AI calls the video tool. It does not scan files or consume resources when idle.

What video formats work?

MP4, MOV, AVI, WebM, MKV, and more. Max 500 MB, 15 minutes for authenticated users.

Do I need Python installed?

Yes. Python 3.9 or later. Install with pip install vidcontext-mcp.

How much does it cost?

The MCP server is free. API usage is $0.20 per minute of video. 20 free credits on signup.

Try it in Cursor

Install in one command, configure in one file. 20 free credits on signup. No credit card required.

Get API key