ScreenMind
Open-source, privacy-first AI screen memory tool that captures and analyzes your screen activity locally to let you search and chat with your screen history.
ScreenMind
What is ScreenMind?
ScreenMind is an AI-powered screen memory application that continuously captures your screen, analyzes it using Gemma 4 (local AI model), and builds a searchable, chat-accessible history of your digital activity. It runs entirely on your machine with zero cloud dependencies.
How to use ScreenMind?
- 1Clone the repository and install dependencies.
- 2Run the application with `python main.py`.
- 3Open `http://127.0.0.1:7777` in your browser.
- 4On first run, download the Gemma 4 model through the Model Hub in the UI.
- 5The tool will start capturing your screen and analyzing it.
- 6Use the dashboard to browse timeline, search, chat, view analytics.
- 7Use hotkeys for instant bookmarks, pause, voice memos.
ScreenMind Key Features
- Smart Capture with content-change detection
- Gemma 4 Vision Analysis (app detection, activity categorization, mood, scene description)
- Hybrid Search (semantic embeddings + FTS5 keyword)
- Chat with Memory (conversational RAG)
- Voice Memos via Gemma 4 audio encoder
- Meeting Transcription with auto-detection and summary
- Analytics Dashboard (category breakdown, top apps, hourly heatmap)
- Day Rewind timelapse playback
- Multi-Model AI Pipeline (EasyOCR, Gemma 4, MiniLM, FTS5)
- Agent Platform (Markdown and Python plugins)
- MCP Server for Claude/Cursor/VS Code
- Integrations: Obsidian, Notion, Webhooks
- Encryption at rest (Fernet + OS keyring)
- Sensitive Data Filter (redacts PII, API keys)
- Incognito Mode
ScreenMind Use Cases
- Personal productivity tracking and analytics
- Automated meeting notes and summaries
- Quick retrieval of past screen information
- Focus and distraction analysis
- Building custom automations on screen data
ScreenMind Pricing & Free Credits
ScreenMind currently operates on a Free model.
This tool is completely free to use
ScreenMind Pros & Cons
Pros
- 100% private and local with zero cloud dependencies
- Open source (MIT license) and completely free
- Runs on any GPU with 4GB+ VRAM, also CPU-only
- Cross-platform (Windows, macOS, Linux)
- Comprehensive features including voice and meeting transcription
- Extensible via agents, MCP, and integrations
Cons
- Requires Python and some setup effort
- Initial model download is ~5GB
- Performance heavily depends on GPU capabilities
- No dedicated commercial support
What is ScreenMind best for?
- Privacy-conscious users who want local screen memory
- Developers building automations on screen data
- Professionals needing searchable screen history
- Users who want local meeting transcription without cloud