screenpipe

Official
TypeScript

Local-first workflow memory for AI agents through MCP/API.

About screenpipe MCP Server

screenpipe captures selected screen, audio, app, and meeting context locally so Claude, Codex, Cursor, and local AI workflows can search real work history through MCP and API. It helps teams turn daily work into cited notes, SOPs, workflow reports, and automation candidates while keeping sensitive context under user control.

Features

Tools
Resources
Getting Started

Install with `npx -y screenpipe-mcp@latest`, make sure the local screenpipe app/API is running, then add the server to Claude Desktop, Cursor, Windsurf, VS Code, or any MCP client.

Tools (2)
Functions this server exposes to AI clients

search

Search local screen, audio, app, and meeting history captured by screenpipe.

query_memory

Retrieve workflow context and cited historical activity for AI agents.

Resources (1)
Data sources this server exposes to AI clients

local workflow memory

Local-first screen, audio, app, and meeting context indexed for MCP clients.

screenpipe://memory/{query}

Example
{"mcpServers":{"screenpipe":{"command":"npx","args":["-y","screenpipe-mcp@latest"]}}}
Compatible With
Claude Desktop
Cursor
Windsurf
VS Code