📌Overview

Purpose: ScreenPipe aims to revolutionize AI application development by providing an open-source framework that records desktop activity locally and in real-time, allowing developers to create context-aware applications.

Overview: ScreenPipe is a desktop application framework that continuously records user interactions with their computer to provide AI developers with rich contextual data. This enables the development of innovative applications tailored to user behavior and preferences, all while maintaining local data privacy.

Key Features:

• 24/7 Local Recording: Continuously captures screen activity and audio, ensuring a comprehensive dataset without reliance on cloud storage, thus enhancing data privacy.

• Easy Plugin Creation: Developers can create custom "pipes" (plugins) using a supported framework, enabling them to build and monetize tailored applications with minimal effort.

---

screenpipe

AI app store powered by 24/7 desktop history
open source | 100% local | dev friendly | 24/7 screen, mic recording

---

[ recording reality, one pixel at a time ]

---

News 🔥

• [2025/03] Introducing screenpipe terminator: playwright but for your desktop. The fastest and most reliable computer use SDK on the internet (100x faster, based on OS APIs instead of vision)
• [2025/02] Hosting a hackathon with $12k in cash prizes, 28 Feb
• [2025/01] Partnering with Different AI to bring you financial automations based on your screen and a drop-in replacement for granola within Obsidian
• [2024/12] Pipe store Stripe integration: devs build cool apps with a few lines of JS and make passive income
• [2024/11] Screenpipe is number 1 GitHub trending repo (again)
• [2024/10] Screenpipe backed by Founders, Inc
• [2024/09] Screenpipe is number 1 GitHub trending repo & featured on HackerNews
• [2024/08] Anyone can now create, share, install pipes (plugins) from the app interface based on a GitHub repo/directory
• [2024/08] Running bounties! Contribute to screenpipe & make money, check issues
• [2024/08] Released Apple & Windows Native OCR
• [2024/07] Launched the desktop app! Download now!

---

How it works?

• Records everything 24/7, 100% locally, uses 10% CPU, 4 GB RAM, 15 GB/month storage
• Indexes recorded data into an API
• Developers build AI apps with the user’s full context, desktop native, Next.js, publish and monetize

---

Why?

• AI models are commoditized
• AI is only as good as its context
• The most valuable context is all contained in your screen

---

Get started

macOS, Linux:

curl -fsSL get.screenpi.pe/cli | sh

Windows:

iwr get.screenpi.pe/cli.ps1 | iex

Then run:

screenpipe

Make sure to allow permissions on macOS (screen, mic)

• Get the desktop app: https://screenpi.pe/
• Docs & build from source: https://docs.screenpi.pe/getting-started

---

Create plugins

screenpipe has a plugin system called "pipe" which lets you create desktop apps in Next.js in a sandboxed environment within Rust code. Read more at: https://docs.screenpi.pe/plugins

To create a pipe:

bunx --bun @screenpipe/dev@latest pipe create

To publish a pipe:

cd foo
bunx --bun @screenpipe/dev@latest pipe register --name foo [--paid --price 50] # subscription option
bun run build
bunx --bun @screenpipe/dev@latest pipe publish --name foo

You can then publish these to the store and monetize.

---

Community

• Template to build screenpipe-powered desktop native app using Tauri: LorenzoBloedow/screenpipe-tauri-template-dev
• Template to build screenpipe-powered desktop native app using Electron: neo773/screenpipe-electron
• Community projects: https://www.sprint.dev/hackathons/screenpipe

---

Star history

---

Contributing

Contributions are welcome! If you'd like to contribute, please read CONTRIBUTING.md.

---

Badges