New Server: Screenpipe - Screen & audio context for AI

[louis030195]

Server Information

Name: Screenpipe MCP Server
Repository: https://github.com/mediar-ai/screenpipe
Description: Provides AI with context from 24/7 screen recordings and audio transcriptions

What it does

Screenpipe MCP server enables AI assistants to:

1. Search screen history - Find text from anything you've seen on screen via OCR
2. Search audio transcriptions - Query meetings, calls, and spoken content
3. Get recent context - Retrieve what's been on screen recently
4. Semantic search - Natural language queries over your digital activity

Why it's useful

This fills a unique gap in the MCP ecosystem - giving AI assistants awareness of the user's visual and audio context. Use cases:

• "What was that error message I saw?"
• "Summarize my last meeting"
• "Find the documentation I was reading about X"
• "What was discussed about feature Y?"

Technical Details

• Language: TypeScript/Rust
• Platform: macOS, Windows, Linux
• Privacy: 100% local, no cloud required
• Stars: 16,500+ on GitHub
• License: MIT

MCP Features

• search_screen - Full-text + semantic search over OCR content
• search_audio - Search audio transcriptions
• get_recent_context - Get recent screen/audio activity
• get_frame - Retrieve specific screenshot

Installation

npm install @screenpipe/mcp

Or via the Screenpipe desktop app which includes the MCP server.

Links

• GitHub: https://github.com/mediar-ai/screenpipe
• Docs: https://docs.screenpi.pe
• MCP Server: https://github.com/mediar-ai/screenpipe/tree/main/screenpipe-js/mcp

[koic]

Since #3046, new entries are no longer added to the README's third-party server list. Going forward, please consider registering it in the MCP Registry. Thank you.