screen-memory-mcp
Enables Claude to remember and search everything that appeared on your screen using local OCR and embeddings. Provides tools for semantic search, activity timeline, and code change tracking.
README
🧠Screen Memory MCP
Give Claude a perfect memory of everything that's ever been on your screen.
"What was I researching on ChatGPT an hour ago?" "What code did I write this morning?" "Walk me through what I did in the last 30 minutes."
Just ask Claude. It remembers.
How it works
Every N seconds:
📸 Screenshot → 🔤 Windows OCR → 🧬 Embed → 💾 SQLite
(built-in, accurate, (local model, (text only,
no hallucination) no API calls) ~5KB/capture)
100% local. No images stored. No cloud. No API costs.
Features
- Semantic search — "find when I was reading about X" across your entire screen history
- Activity timeline — connected narrative of what you did and why, not just isolated snapshots
- Code tracking — detects VS Code / PyCharm, tracks which files you edited, computes diffs between captures
- App awareness — knows if you were in a browser, editor, or terminal and tags captures accordingly
- Tiny footprint — text + embeddings only, ~5KB per capture, ~30MB/day at 10s interval
Requirements
- Windows 10 / 11
- Python 3.11+
- No Tesseract. No cloud APIs. No external accounts.
Installation
git clone https://github.com/YOUR_USERNAME/screen-memory-mcp
cd screen-memory-mcp
pip install -r requirements.txt
Usage
1. Start the capture daemon
python -m screen_memory.capture --interval 10
| Interval | Storage/day | Recommended for |
|---|---|---|
| 5s | ~90MB | Maximum detail |
| 10s | ~45MB | Daily use ✓ |
| 30s | ~15MB | Light use |
2. Connect to Claude Desktop
Edit %APPDATA%\Claude\claude_desktop_config.json:
{
"mcpServers": {
"screen-memory": {
"command": "C:\\Python314\\python.exe",
"args": ["-m", "screen_memory.server"],
"cwd": "C:\\path\\to\\screen-memory-mcp",
"env": {
"PYTHONPATH": "C:\\path\\to\\screen-memory-mcp"
}
}
}
}
Restart Claude Desktop. The screen-memory tools will appear automatically.
3. Ask Claude anything
"What was I doing 20 minutes ago?"
"Summarize my activity for the last hour"
"What code did I write today in VS Code?"
"Find when I was reading about async Python"
"How did server.py change during my coding session?"
MCP Tools
| Tool | Description |
|---|---|
search_screen_memory |
Semantic search across all captures |
get_activity_timeline |
Chronological narrative of a time range |
get_code_changes |
Code diffs per file for a coding session |
get_recent_captures |
Latest N captures |
get_capture |
Full text of a specific capture by ID |
get_stats |
DB size, capture count, date range |
Auto-start on login (Windows)
Run once in an elevated terminal:
schtasks /create /tn "ScreenMemoryCapture" /tr "C:\path\to\start_capture.bat" /sc onlogon /ru YOUR_USERNAME /f
Storage
All data lives in ~/.screen_memory/memory.db — a single SQLite file.
To wipe everything: just delete it.
Privacy
- ✅ Fully offline after install
- ✅ No screenshots saved to disk
- ✅ All text stays on your machine
- ✅ Open source — every line auditable
- ✅ Delete anytime:
rm -rf ~/.screen_memory
Roadmap
- [ ] macOS / Linux support
- [ ] Auto-cleanup (configurable retention period)
- [ ] System tray app (start/stop without terminal)
- [ ] Screenshot viewer (opt-in)
- [ ] Multi-monitor support
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.