OSS TanbouOn this page
Overview
screenpipe continuously captures screen activity, accessibility data, OCR fallbacks, and audio into local searchable storage. Its REST API and MCP integrations let agents such as Claude and Codex query recent computer history as working context.
Features and best fit
Based on official documentation; not hands-on tested · Content checked:
Key features
Capture meaningful screen changes with accessibility data and OCR fallback
screenpipe uses event-driven capture around meaningful activity, pairs screenshots with accessibility information when available, and falls back to OCR when necessary.
Sources: [2]
Search local screen history and audio transcripts
System audio and microphone input can be transcribed and stored alongside visual history, with local SQLite and REST access for searching past work and conversations.
Sources: [2]
Expose computer history to coding agents through CLI and MCP
The CLI provides record and setup flows, and MCP integrations let Claude, Codex, Cursor, and other agents query recent context.
Sources: [2]
Best fit
Fits local-first personal AI memory across meetings, browsers, and IDEs
It is useful when work spans multiple applications and the user wants searchable recall or recent desktop context available to an AI agent.
Sources: [2]
Before adoption
Current source is source-available and commercial use requires a separate license
LICENSE.md permits personal non-commercial use, non-profit, educational and research use, and limited organizational evaluation, while defining commercial use as requiring a paid commercial license.
Sources: [4]
Estimate storage and device cost for continuous capture
The README recommends 8 GB RAM and gives roughly 5–10 GB per month as a storage guideline. Multi-monitor and long-running audio usage should be evaluated against retention and disk capacity.
Sources: [2]
Review data flow when enabling optional cloud features
Core data is local by default, while optional cloud AI and encrypted sync features also exist. For sensitive material, review capture exclusions, selected models, and synchronization destinations.
Sources: [2]
Official sources
- [1]screenpipe/screenpipe repository(2026-10-01)
- [2]screenpipe README(2026-10-01)
- [3]Screenpipe App v2.7.79 release(2026-10-01)
- [4]Screenpipe Commercial License(2026-10-01)
Supplemental curator note
screenpipe builds a local-first computer history that agents such as Codex and Claude can query for recent visual and conversational context. The current code is source-available under a Commercial License rather than an OSI-style open-source license, so business and commercial-use terms need explicit review.
Try it in 3 steps
- 1
Start local recording
Use the documented CLI path to begin local screen and audio capture. Review the capture scope because recorded activity can include sensitive information.
npx screenpipe record - 2
Configure agent integrations
Install the screenpipe skill and MCP configuration into supported agents detected on the computer.
npx screenpipe setup - 3
Verify history search
Use the README example to confirm that recent computer history can be retrieved through the configured agent.
Ask your configured agent: what did I see in the last 5 mins?
Growth
Growth trends · Last 30 days
21,797 Stars
Trend data is still being collected.
Development activity
Last 90 days · weekly
- Commits (last 30 days)
- 619
- Open PRs
- 21
Development activity is still being collected.
Built with
Categories and tags
GitHub data
GitHub dataView detailed GitHub data
GitHub Topics
- ai
- computer-vision
- llm
- machine-learning
- multimodal
- agents
- agi
- audio-recording
- local-ai
- local-first
- privacy
- screen-recording
- Stars
- 21,797
- Forks
- 2,226
- Watchers
- 117
- Open issues
- 12
- Contributors
- 158
- Owner type
- Organization
- Primary language
- Rust
- License
- Not determined
- Repository last updated
- Oct 2, 2026
Related information
Write a related articleShare a guide or use case for this OSS in Markdown. Articles are published after administrator approval.
Explore next
- Tolaria19,951 Stars
3 shared tag(s) · 1 shared category(s)
Use Markdown and Git as the source of truth for a knowledge base shared with AI agents, MCP, and local or hosted models
TypeScript - Synara1,927 Stars
3 shared tag(s) · 1 shared category(s)
Unify Codex, Claude Code, and other provider sessions with worktrees, diffs, terminals, browser previews, and Git review in one local-first workspace
TypeScript - Tauri111,529 Stars
3 shared tag(s) · Same language
Combine a web frontend and Rust native layer into desktop and mobile apps using system WebViews
Rust - Handy32,617 Stars
4 shared tag(s) · 2 shared category(s) · Same language
run Whisper or Parakeet on-device and dictate into any desktop application from a global shortcut
Rust - Meetily31,356 Stars
4 shared tag(s) · 2 shared category(s) · Same language
transcribe meetings on-device with Whisper or Parakeet and choose local or cloud providers for summarization
Rust - Minutes1,523 Stars
4 shared tag(s) · 2 shared category(s) · Same language
Transcribe meetings, calls, and voice memos on-device into Markdown and selectively reuse them through MCP, CLI, and AI agents
Rust
Report incorrect information
Tell us if any listing information is incorrect or outdated.