How mediar-ai/screenpipe works
Main Components Identification
a) Core Recording Components:
• screenpipe-vision (Screen capture and OCR)
• screenpipe-audio (Audio recording and processing)
• screenpipe-actions (Keyboard/mouse monitoring)
b) Data Processing & Storage:
• screenpipe-server (Central server component)
• SQLite database (Local data storage)
• Video/Audio cache system
c) Plugin System:
• Pipe framework
• Plugin sandbox environment
• Plugin store integration
d) User Interface:
• Tauri-based desktop app (screenpipe-app-tauri)
• Web-based UI components
• System tray integration
┌─ User Interface Layer ─────────────────────┐
• Desktop App (Tauri) │
• Web UI Components │
• System Tray │
• Plugin Layer ────────────────────────────┤
• Plugin Manager │
• Plugin Store │
• Plugin Sandbox │
• Core Services Layer ────────────────────┤
• Vision Service │
• Audio Service │
• Action Service │
• Processing Layer ─────────────────────────┤
• OCR Processing │
• Speech-to-Text │
• Speaker Identification │
• Storage Layer ──────────────────────────┤
• SQLite Database │
• Video Cache │
• Audio Cache │
• ┘
Data Flow Representation
• Screen/audio capture flow → processing → storage
• Plugin data access paths
• User interface interactions
• Real-time processing pipelines
Integration Points
• API endpoints
• Plugin extension points
• External service connections (e.g., AI services)
Monitoring & Control
• Health check flows
• Resource monitoring paths
• System control points
Data security boundaries
This system design should reflect Screenpipe's core value proposition of continuous recording while highlighting its extensible architecture through the plugin system.