mcp-registry/screenpipe

    ==================
      
       /// MCP ///
      /// SCR ///
        
    ==================
        
    [server:online]
    [protocol:ready]

screenpipe

by mediar-ai

AI app-store platform that continuously records a user’s desktop (screen + mic) locally, indexes it, and exposes an API so developers can build context-aware AI desktop apps (“pipes”) in Next.js, publish them and monetise through the built-in store.

15.4k
1.2k
Open Source
GitHubWebsiteDocumentationNPM

01

screenpipe

Command-line entry point that starts the ScreenPipe desktop recorder and local API.

02

pipe create

Generate a new ScreenPipe plugin ("pipe") project scaffold.

03

pipe register

Register a plugin in the ScreenPipe store, optionally setting pricing information.

04

pipe publish

Publish a built plugin to the ScreenPipe store so users can install it.

05

screenpipe terminator

Desktop-automation SDK ("Playwright for your desktop") providing fast, OS-level control of the computer.

Installation

1. Prerequisites
• Node.js ≥ 20 and pnpm (recommended) or npm.
• FFmpeg available in PATH (used for muxing screen & audio).
• macOS / Windows: grant screen-capture & microphone permissions.
2. Clone & install
git clone https://github.com/mediar-ai/screenpipe.git cd screenpipe pnpm install # or: npm install
3. Build native modules & client
pnpm run build # transpile TypeScript → JS pnpm run electron:build # package desktop recorder (optional)
4. Run the MCP server locally
# Starts recorder + MCP REST/WS endpoints on http://localhost:6399 pnpm start
5. Docker (headless servers)
docker build -t screenpipe . docker run --privileged \ -e SCREENPIPE_PORT=6399 \ -v ~/.screenpipe:/data \ screenpipe
6. Configuration (edit `.env` or supply env-vars)
SCREENPIPE_PORT=6399
SCREENPIPE_DATA_DIR=~/.screenpipe
OPENAI_API_KEY=sk-... # if using cloud LLMs
EMBEDDING_MODEL=local/ggml # alternatively use local models
7. Access the web UI
Open http://localhost:6399 in a browser; install/enable extensions or agent plug-ins as prompted.

Documentation

License: MIT License
Updated 7/30/2025

Table of Contents