Pywellen Mcp

PyWellen MCP - Waveform Analysis via Model Context Protocol

[](https://github.com/fvutils/pywellen-mcp/actions) [](https://fvutils.github.io/pywellen-mcp) [](https://www.python.org/downloads/) [](https://opensource.org/licenses/Apache-2.0)

PyWellen MCP is a powerful Model Context Protocol (MCP) server that enables LLM agents to interact with digital waveform files. Analyze VCD, FST, GHW, and other waveform formats using natural language queries and AI-powered tools.

✨ Features

• 🎯 35+ MCP Tools across 9 comprehensive categories
• 📊 Multi-format Support: VCD, FST, GHW, LXT, LXT2, VZT waveforms
• 🔍 Natural Language Queries: Ask about signals using plain English
• ⚡ High Performance: Multi-threaded parsing, LRU caching, optimized algorithms
• 🔗 External Integrations: GTKWave, Verdi, Simvision viewer support
• 📤 Export Capabilities: CSV, JSON, YAML, hierarchy trees, signal lists
• 🧠 LLM Optimization: Signal summarization, pattern detection, recommendation
• 🔒 Production Ready: Comprehensive error handling, security, monitoring

🚀 Quick Start

Installation

# From PyPI (when published)
pip install pywellen-mcp

# From source
git clone https://github.com/fvutils/pywellen-mcp.git
cd pywellen-mcp
pip install -e ".[dev]"

Configuration

Add to your MCP client configuration (e.g., Claude Desktop):

{
  "mcpServers": {
    "pywellen": {
      "command": "pywellen-mcp",
      "args": []
    }
  }
}

Example Usage

# Chat with your LLM using natural language:
"Open the waveform file /path/to/design.vcd"
"Show me all clock signals"
"What's the value of top.cpu.reset at time 1000?"
"Compare signals clk_a and clk_b"
"Export the signal data to CSV"

📊 Current Status

35 Tools Implemented | 182/193 Tests Passing | 94.3% Success Rate

• ✅ Phase 1: Core Infrastructure (4 tools)
• ✅ Phase 2: Hierarchy Navigation (4 tools)
• ✅ Phase 3: Signal Data Access (5 tools)
• ✅ Phase 4: Debugging & Analysis (7 tools)
• ✅ Phase 5: Comparison & Format Conversion (7 tools)
• ✅ Phase 6: LLM Optimization (5 tools)
• ✅ Phase 7: Export & Integration (8 tools)
• 🚧 Phase 8: Production Readiness (CI/CD, security, monitoring)

🛠️ Tool Categories

Core Operations (4 tools)

• waveform_open - Open waveform files (VCD, FST, GHW)
• waveform_close - Close sessions
• waveform_info - Get waveform metadata
• waveform_list_sessions - List active sessions

Hierarchy Navigation (4 tools)

• hierarchy_list_top_scopes - List top-level design scopes
• hierarchy_get_scope - Get scope details
• hierarchy_list_variables - List variables in a scope
• hierarchy_search - Search hierarchy with patterns

Signal Analysis (5 tools)

• signal_get_value - Get signal value at specific time
• signal_get_values - Get values over time range
• signal_get_changes - Get value change events
• signal_get_statistics - Calculate signal statistics
• signal_search - Search signals with filters

Time Management (2 tools)

• time_get_range - Get simulation time range
• time_convert - Convert time units

Debugging & Analysis (7 tools)

• debug_find_transitions - Find signal transitions
• debug_trace_causality - Trace signal causality
• debug_compare_waveforms - Compare waveforms
• debug_build_timeline - Build event timelines
• debug_check_protocol - Protocol checker
• debug_identify_glitches - Glitch detection
• debug_find_correlation - Signal correlation

Comparison (3 tools)

• compare_signals - Compare signal values
• compare_waveforms - Compare entire waveforms
• compare_time_ranges - Compare time ranges

Format Conversion (4 tools)

• format_value - Format signal values
• format_as_signed - Convert to signed values
• format_as_binary - Binary representation
• format_as_hex - Hex representation

LLM Optimization (5 tools)

• query_natural_language - Natural language queries
• signal_summarize - Automatic signal summarization
• recommend_related_signals - Signal recommendations
• docs_get_started - Getting started guide
• docs_tool_guide - Tool usage documentation

Export & Integration (8 tools)

• export_to_csv - Export signals to CSV
• export_hierarchy_tree - Export design hierarchy
• load_signal_list - Load signal configurations
• save_signal_list - Save signal configurations
• export_signal_data - Export to JSON/YAML
• integration_launch_viewer - Launch external viewers
• integration_watch_file - File change monitoring
• integration_generate_gtkwave_save - Generate GTKWave saves

📖 Documentation

• [Getting Started](docs/getting_started.rst) - Installation and first steps
• [Quick Reference](docs/quick_reference.rst) - Common workflows
• [API Reference](docs/api_reference.rst) - Complete tool documentation
• [Best Practices](docs/best_practices.rst) - Performance and optimization
• [Deployment Guide](docs/deployment.rst) - Production deployment
• Full Documentation - Complete docs

🎯 Use Cases

For Verification Engineers

• Analyze waveforms without leaving your LLM chat
• Natural language debugging: "Show me when reset goes low"
• Automated signal correlation workflows
• Quick protocol compliance checks

For Hardware Designers

• Interactive design exploration
• Compare pre/post synthesis waveforms
• Generate test reports automatically
• Integration with existing EDA tools

For Tool Developers

• MCP-based waveform analysis API
• Extensible plugin architecture
• Support for custom waveform formats
• Python-based scripting interface

🔧 Advanced Features

Performance Optimization

• Multi-threaded VCD parsing for faster file loading
• LRU caching for frequently accessed signals
• Lazy loading of signal data on demand
• Efficient time range queries with binary search

Security

• Path validation prevents directory traversal
• Command injection protection for viewer launches
• File permission checks before operations
• Session isolation prevents cross-session access

Error Handling

• Structured error responses with context
• Recovery strategies for common failures
• Graceful degradation on missing data
• Detailed logging for debugging

🧪 Development

Running Tests

# Run all tests
pytest

# Run with coverage
pytest --cov=pywellen_mcp --cov-report=html

# Run specific category
pytest tests/unit/test_tools_llm.py
pytest tests/unit/test_tools_export.py

Performance Benchmarking

# Run benchmark suite
python scripts/benchmark.py

# Profile specific operations
python -m cProfile -s cumtime scripts/benchmark.py

Security Audit

# Run security checks
python scripts/security_audit.py

# Check specific categories
python scripts/security_audit.py --check-paths
python scripts/security_audit.py --check-commands

🏗️ Architecture

Components

pywellen-mcp/
├── src/pywellen_mcp/
│   ├── server.py              # MCP server implementation
│   ├── session.py             # Session management
│   ├── tools_waveform.py      # Core waveform operations
│   ├── tools_hierarchy.py     # Hierarchy navigation
│   ├── tools_signal.py        # Signal analysis
│   ├── tools_time.py          # Time management
│   ├── tools_debug.py         # Debugging tools
│   ├── tools_compare.py       # Comparison operations
│   ├── tools_format.py        # Format conversion
│   ├── tools_llm.py           # LLM optimization
│   ├── tools_export.py        # Export capabilities
│   └── tools_integration.py   # External integrations
├── tests/
│   └── unit/                  # Comprehensive unit tests
├── scripts/
│   ├── benchmark.py           # Performance benchmarks
│   └── security_audit.py      # Security checks
└── docs/                      # Sphinx documentation

Session Lifecycle

1. Open: waveform_open creates session with unique ID
2. Use: Tools access session via session_id parameter
3. Cleanup: Automatic after 1 hour timeout or explicit close

Error Handling

All operations return structured errors:

{
  "error": "SESSION_NOT_FOUND",
  "message": "Session abc123 not found",
  "context": {
    "session_id": "abc123",
    "active_sessions": ["def456"]
  }
}

🤝 Contributing

We welcome contributions! Please see our [Contributing Guide](docs/contributing.rst).

Development Setup

# Clone repository
git clone https://github.com/fvutils/pywellen-mcp.git
cd pywellen-mcp

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install in development mode
pip install -e ".[dev]"

# Run tests
pytest

Code Standards

• Style: Black formatting, PEP 8 compliance
• Type Hints: Full type annotations
• Documentation: Docstrings for all public APIs
• Testing: Minimum 80% code coverage

📝 Roadmap

Phase 8: Production Readiness (In Progress)

• [x] CI/CD pipeline (GitHub Actions)
• [x] Performance benchmarking suite
• [x] Security audit script
• [x] Comprehensive documentation
• [ ] Memory profiling
• [ ] Integration tests with real waveforms
• [ ] Version 1.0.0 release

Future Enhancements

• [ ] WebSocket-based streaming for large waveforms
• [ ] Distributed analysis for massive designs
• [ ] Machine learning-based anomaly detection
• [ ] Plugin system for custom analyzers
• [ ] Support for SystemVerilog assertions
• [ ] Real-time waveform monitoring

🙏 Acknowledgments

• Wellen - Rust waveform parsing library
• MCP - Model Context Protocol specification
• Anthropic - MCP development and Claude integration
• GTKWave, Verdi, Simvision - Waveform viewer integrations

📄 License

Licensed under the Apache License, Version 2.0. See [LICENSE](LICENSE) for details.

🔗 Links

• Homepage: https://github.com/fvutils/pywellen-mcp
• Documentation: https://fvutils.github.io/pywellen-mcp
• PyPI: https://pypi.org/project/pywellen-mcp/
• Issues: https://github.com/fvutils/pywellen-mcp/issues
• Discussions: https://github.com/fvutils/pywellen-mcp/discussions

📧 Contact

• Author: Matthew Ballance
• Email: mballance@fvutils.com
• GitHub: @mballance

---

Made with ❤️ by the FVUtils community

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: fvutils
• Source: fvutils/pywellen-mcp
• License: Apache-2.0
• Homepage: https://fvutils.github.io/pywellen-mcp/

Install and usage instructions live in the source repository linked above.