
qudi MCP Integration
Enables natural language control of quantum photonics experiments through the qudi framework. Supports safe instrument control, measurement execution, and safety system management with built-in runlevel protection and parameter validation.
README
🔬 qudi MCP Integration
Model Context Protocol (MCP) integration for qudi quantum photonics experiments
Enable natural language control of quantum experiments through Claude Desktop!
🚀 Quick Start
1. Installation Options
📋 Standalone Mode (Recommended for Testing)
# Clone this repository
git clone https://github.com/dirkenglund/qudi-mcp-integration.git
cd qudi-mcp-integration
# Install minimal dependencies (simulation only)
pip install -r requirements-standalone.txt
🖼️ With Plot Extraction Capabilities
# Install standalone + plot extraction
pip install -r requirements-standalone.txt
pip install -r requirements-plot-extraction.txt
🔬 Full qudi Integration (For Hardware Control)
# For real quantum hardware control
pip install -r requirements-full.txt
# Additional setup required:
# 1. Install qudi-core separately
# 2. Configure hardware drivers
# 3. Set up measurement modules
2. Configure Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json
:
{
"mcpServers": {
"qudi-mcp": {
"command": "/opt/homebrew/bin/python3",
"args": ["/path/to/qudi-mcp-integration/simple_mcp_server.py"],
"env": {
"PYTHONPATH": "/path/to/qudi-mcp-integration",
"QUDI_MCP_LOG_LEVEL": "INFO"
}
}
}
}
3. Test with Claude Desktop
Restart Claude Desktop and try:
"Get qudi station information"
Example Commands:
- "Start a photoluminescence scan from 630-650 nm"
- "Check all safety interlocks before measurement"
- "Can I safely set laser power to 5 milliwatts?"
Features
🛡️ Safety First
- Runlevels:
dry-run
→sim
→live
progression - Parameter Validation: All values checked against safety limits
- Emergency Stop: Immediate halt capability
- Interlocks: Critical system monitoring
🔧 Instrument Control
- List and load qudi instruments
- Get/set parameters with safety validation
- Real-time status monitoring
- Simulated operation for safe testing
📊 Measurement Execution
- Available modules: PL scan, gate sweep, resonance scan, time trace, 2D maps
- Progress monitoring and data acquisition
- Configurable measurement parameters
- Data export capabilities
🤖 LLM Integration
- Natural language command processing
- Contextual tool suggestions
- Comprehensive error reporting
- Audit logging for all operations
🖼️ Plot Extraction (Optional)
- RKHS Spline Projection - Mathematical smoothing using reproducing kernel Hilbert spaces
- Computer Vision - Extract data points from scientific plots and graphs
- Spectrum Analysis - Advanced processing for spectroscopy data
- Multiple Formats - Support for PNG, JPG, TIFF, and other image formats
- Quantum Data - Optimized for photoluminescence and transport measurements
Plot Extraction Commands:
- "Extract data from this plot image: /path/to/spectrum.png"
- "Analyze spectrum with RKHS smoothing using epsilon 0.05"
- "What plot extraction capabilities are available?"
Architecture
Claude ←→ MCP ←→ qudi_mcp_server ←→ qudi ←→ Instruments
↓
Safety System
Runlevel Control
Parameter Validation
Core Components
qudi_mcp_server.py
: Main MCP server and tool routersafety.py
: Safety system and runlevel managementtools/
: Tool implementations (instruments, measurements, safety)claude_config/
: Claude Desktop configuration templates
Safety System
Runlevels
dry-run
(default): Simulation only, no hardware interactionsim
: Realistic simulation with hardware-like responseslive
: Real hardware control (requires approval + safety checks)
Built-in Limits
- Laser power: 0-10 mW
- Gate voltages: ±2.0 V
- Bias voltages: ±1.0 V
- Temperature: 0.01-300 K
- Magnetic field: ±9.0 T
- Measurement time: 0.001-3600 s
Emergency Procedures
All emergency stop triggers:
- Halt all running measurements
- Force runlevel to
dry-run
- Log incident with timestamp
- Require manual reset with confirmation
Usage Examples
System Status
# Check system status
→ Get qudi station information
# Response: runlevel, loaded instruments, active measurements, safety status
Instrument Control
# List instruments
→ List available qudi instruments
# Load an instrument
→ Load the laser_controller instrument
# Set parameter safely
→ Set laser power to 2.5 mW
Measurements
# Start a measurement
→ Start a photoluminescence scan from 630 to 650 nm with 0.5 second integration
# Check progress
→ What's the status of running measurements?
# Get results
→ Get the measurement data for the PL scan
Safety Operations
# Check safety systems
→ Check all safety interlocks
# Change runlevel (when ready for hardware)
→ Set runlevel to sim mode for realistic testing
# Emergency procedures
→ Emergency stop all operations
Development Status
✅ Completed (Phase 1)
- MCP server framework
- Safety system with runlevels and limits
- Tool architecture for instruments, measurements, safety
- Claude Desktop integration
- Comprehensive simulation mode
- Documentation and setup guides
🚧 In Progress (Phase 2)
- qudi core integration
- Real instrument driver connections
- Hardware abstraction layer
- Live mode approval workflows
📋 Planned (Phase 3)
- Advanced measurement protocols
- Data analysis tool integration
- Multi-user access control
- Web-based monitoring interface
File Structure
mcp_integration/
├── __init__.py # Package initialization
├── qudi_mcp_server.py # Main MCP server
├── safety.py # Safety and runlevel system
├── tools/ # MCP tool implementations
│ ├── __init__.py
│ ├── instrument_tools.py # Instrument control tools
│ ├── measurement_tools.py # Measurement execution tools
│ └── safety_tools.py # Safety and emergency tools
├── claude_config/ # Claude configuration templates
├── README.md # This file
└── requirements.txt # Python dependencies
Development Guidelines
Adding New Tools
- Implement in appropriate
tools/
module - Register in
qudi_mcp_server.py
tool list - Add safety validation for parameters
- Test thoroughly in dry-run mode
- Document in tool docstrings
Safety Requirements
- All write operations must validate parameters
- Critical operations need explicit approval in live mode
- Comprehensive error handling and logging required
- Emergency stop must work from any state
Testing Protocol
- Dry-run: Logic validation without hardware
- Simulation: Realistic behavior testing
- Hardware: Real instrument validation (when available)
- Safety: Verify all safety mechanisms
- Integration: End-to-end workflow testing
Troubleshooting
Common Issues
"MCP package not found"
pip install mcp
"Tool not found" errors
- Check tool registration in
qudi_mcp_server.py
- Verify tool implementation in
tools/
modules
Safety validation failures
- Check parameter values against limits in
safety.py
- Use
safety.get_limits
to see current constraints
Claude Desktop not seeing tools
- Verify absolute paths in configuration file
- Restart Claude Desktop completely
- Check Python path and MCP server execution
Getting Help
- Check logs: MCP server logs to stderr
- Test tools directly: Run
python qudi_mcp_server.py
- Validate config: Check Claude Desktop config file syntax
- Start simple: Begin with
station.info
andsafety.check_interlocks
Contributing
This integration is part of the MIT QPG development branch. To contribute:
- Fork the repository
- Create feature branches from
dev/llm-mcp-automation
- Follow safety-first development practices
- Include comprehensive tests
- Update documentation for new features
Repository: https://github.com/dirkenglund/qudi-iqo-modules-QPG
Branch: dev/llm-mcp-automation
Documentation: See docs/LLM_MCP_INTEGRATION.md
for full details
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.