screenpipe/screenpipe

What it solves

Screenpipe creates a searchable, AI-powered memory of your computer activity. It solves the problem of forgetting what you saw, heard, or did across various applications, providing a local-first way to recall information and give AI agents context about your actual workflow.

How it works

The system uses event-driven capture to record screenshots and the OS accessibility tree (falling back to OCR) whenever meaningful activity occurs (e.g., app switches, clicks, typing). It also captures system and microphone audio, transcribing it locally using Whisper. All data is stored in a local SQLite database with full-text search. It exposes a REST API and an MCP (Model Context Protocol) server, allowing AI assistants like Claude or Cursor to query your history.

Who it’s for

• Knowledge workers and researchers who need to recall specific details from their screen or meetings.
• Developers who want to provide AI coding assistants with real-time context of their work.
• People with ADHD who struggle to keep track of documents and tabs.
• Teams and enterprises looking to deploy AI agents with deterministic, OS-level data permissions.

Highlights

• 100% Local Storage: Data stays on your machine by default, ensuring privacy.
• Pipes: A plugin system where scheduled AI agents are defined as markdown files to automate tasks like meeting summaries or day recaps.
• MCP Server: Native integration with AI assistants (Claude, Cursor, etc.) to query screen history.
• Event-Driven Capture: Reduces CPU and storage overhead by only recording when changes occur.
• Multi-Monitor Support: Captures all connected screens simultaneously.