Screenpipe logo

Screenpipe

24/7 local screen and audio capture with AI-powered search

Record your desktop and microphone continuously, store data locally, and retrieve moments instantly using natural-language AI queries—all without internet connectivity.

Screenpipe banner

Overview

Screenpipe continuously records your screen and microphone, storing the media on your own machine. By keeping everything 100 % local, it guarantees privacy while providing a searchable replay of your digital activity.

Capabilities

The captured footage can be queried with natural-language prompts, allowing you to locate specific moments, conversations, or visual states instantly. Its lightweight design runs at roughly 10 % CPU usage, consumes about 4 GB RAM, and generates around 15 GB of storage per month, making it suitable for everyday desktop use.

Deployment

Installation is a single command on macOS, Linux, or Windows, with an optional desktop UI for visual control. All processing happens offline, and the system can be extended via APIs to integrate custom AI models or downstream workflows.

Highlights

Continuous 24/7 screen and microphone recording
Fully local storage; data never leaves the machine
Natural-language AI search across captured media
Low resource footprint (≈10 % CPU, 4 GB RAM, ~15 GB/month)

Pros

  • Complete privacy; no cloud upload
  • Immediate access to visual and audio history
  • Works offline; no internet required
  • Simple cross-platform installation

Considerations

  • Storage grows ~15 GB per month; requires disk management
  • No built-in remote sharing or collaboration features
  • AI search depends on external models you must provide
  • High-frequency recording may impact battery on laptops

Fit guide

Great for

  • Developers building AI agents that need visual context
  • Researchers analyzing user interaction over time
  • Privacy-focused power users who want a personal replay archive
  • Content creators reviewing past recordings without cloud services

Not ideal when

  • Users with limited disk space or strict storage quotas
  • Teams needing shared, synchronized recordings
  • Environments where continuous audio capture is prohibited
  • Mobile devices where battery life is critical

How teams use it

Debugging UI issues

Capture every screen change, then query the exact moment an error appeared, speeding up troubleshooting.

Training multimodal AI models

Provide a locally stored dataset of synchronized video and audio for model fine-tuning without exposing proprietary content.

Personal knowledge base

Search past webinars or meetings by speaking a keyword, retrieving the exact clip instantly.

Compliance auditing

Maintain an immutable local log of screen activity for internal audits while keeping data under full control.

Tech snapshot

TypeScript75%
Rust22%
MDX1%
JavaScript1%
CSS1%
Shell1%

Tags

mlvisionaicomputer-visionllmmachine-learningagentsagimultimodal

Frequently asked questions

Does screenpipe upload any data to the cloud?

No. All recordings are stored on the local machine and never leave unless you manually export them.

What operating systems are supported?

Pre-built binaries are available for macOS, Linux, and Windows.

How much storage will it use?

Approximately 15 GB per month of video and audio, depending on screen resolution and activity.

Can I integrate my own AI model for search?

Yes. screenpipe provides an API that lets you plug in any local LLM or embedding model.

Is there a graphical interface?

A desktop app is offered for visual control; the CLI can also be used for scripting.

Project at a glance

Active
Stars
17,125
Watchers
17,125
Forks
1,439
Repo age1 year old
Last commit2 days ago
Primary languageRust

Last synced 2 days ago