README.md•7.38 kB
# React Native Godot Documentation MCP Server
A powerful Model Context Protocol (MCP) server for intelligent access to React Native Godot documentation, examples, and implementation guides. This server enables LLMs to efficiently fetch and search documentation from the [react-native-godot](https://github.com/borndotcom/react-native-godot) repository by Born.com and Migeran.
## 🚀 Features
### Agent-Centric Design
This MCP server follows best practices for agent-oriented tool design:
- **Workflow-focused tools** that enable complete tasks, not just API wrapping
- **Optimized for context efficiency** with configurable detail levels
- **Actionable error messages** that guide agents toward solutions
- **Natural task subdivisions** with intuitive tool naming
### Available Tools
#### 📖 `get_documentation`
Fetch specific sections of React Native Godot documentation:
- Overview, installation, initialization
- API usage, threading, views
- Export, debugging, custom builds
- Configurable detail levels (concise/detailed/full)
#### 🔍 `search_documentation`
Intelligent search across all documentation:
- Keyword-based search with relevance scoring
- Returns most relevant sections
- Perfect for finding specific topics or troubleshooting
#### 💻 `get_example_code`
Retrieve working code examples for:
- Initialization and setup
- API usage patterns
- Signal handling
- View embedding
- Worklets and threading
- Complete application examples
#### 🛠️ `get_setup_instructions`
Platform-specific setup guidance:
- iOS and Android configurations
- Debugging setup options
- Custom build instructions
- Step-by-step installation process
#### 📚 `get_api_reference`
Detailed API documentation for:
- RTNGodot class methods
- RTNGodotView component
- runOnGodotThread function
- Signals and callables
- Property access patterns
#### 🔧 `get_troubleshooting`
Solutions for common issues:
- Build errors
- View display problems
- Threading issues
- Export problems
- Performance optimization
#### 📁 `get_file_from_repo`
Direct access to repository files:
- Example implementations
- Configuration files
- Build scripts
- Native code
## 📦 Installation
### Prerequisites
```bash
# Python 3.8+
python --version
# Install MCP and FastMCP
pip install mcp fastmcp
```
### Install Dependencies
```bash
pip install httpx pydantic
```
### Make Executable
```bash
chmod +x react_native_godot_mcp.py
```
## 🎯 Usage
### Running the Server
#### Standalone Mode
```bash
python react_native_godot_mcp.py
```
#### With MCP Inspector (for testing)
```bash
npx @modelcontextprotocol/inspector python react_native_godot_mcp.py
```
### Integration with Claude Desktop
Add to your Claude configuration file:
**MacOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
**Windows**: `%APPDATA%/Claude/claude_desktop_config.json`
```json
{
"mcpServers": {
"react-native-godot": {
"command": "python",
"args": ["/path/to/react_native_godot_mcp.py"],
"env": {}
}
}
}
```
### Integration with Other MCP Clients
The server uses stdio transport and can be integrated with any MCP-compatible client:
```python
# Example Python client usage
from mcp import Client
client = Client()
client.connect_stdio(["python", "react_native_godot_mcp.py"])
# Use tools
result = await client.call_tool(
"get_documentation",
{"section": "initialization", "detail": "detailed"}
)
```
## 💡 Example Queries
### Get Started with React Native Godot
```python
# Fetch installation instructions
get_setup_instructions(platform="both", include_debugging=True)
# Get initialization example
get_example_code(topic="initialization", platform="ios")
```
### Search for Specific Topics
```python
# Search for worklet information
search_documentation(query="worklets threading", max_results=5)
# Find signal handling docs
search_documentation(query="connect signals JavaScript")
```
### Troubleshoot Issues
```python
# Get help with build errors
get_troubleshooting(issue="build_error", platform="android")
# Debug view problems
get_troubleshooting(issue="view_not_showing")
```
### Deep Dive into API
```python
# Get RTNGodot API reference
get_api_reference(topic="RTNGodot", include_examples=True)
# Learn about signals
get_api_reference(topic="signals")
```
## 🧪 Running Evaluations
The repository includes comprehensive evaluation questions to test the MCP server's capabilities.
### Using the Evaluation File
1. The evaluation file (`react_native_godot_evaluation.xml`) contains 10 complex questions
2. Each question tests the server's ability to find specific information
3. Questions cover initialization, threading, debugging, API usage, and more
### Running Evaluations with MCP
```bash
# Install evaluation tools
pip install mcp-eval
# Run evaluation
mcp-eval run react_native_godot_mcp.py react_native_godot_evaluation.xml
```
### Manual Testing
You can manually test each tool:
```python
# Test documentation fetching
python -c "
import asyncio
from react_native_godot_mcp import get_documentation
async def test():
result = await get_documentation(None, 'threading', 'markdown', 'detailed')
print(result)
asyncio.run(test())
"
```
## 🏗️ Architecture
### Design Principles
1. **Workflow-Oriented**: Tools are designed around common developer workflows
2. **Context-Aware**: Adjustable detail levels to optimize token usage
3. **Error-Resilient**: Graceful handling with actionable error messages
4. **Format-Flexible**: Supports both Markdown and JSON responses
### Response Formats
All tools support two response formats:
- **Markdown**: Human-readable, formatted documentation
- **JSON**: Structured data for programmatic use
### Detail Levels
Three levels of detail for documentation:
- **Concise**: Key points and headers only
- **Detailed**: Standard documentation with examples
- **Full**: Complete content including all details
## 🔒 Security & Limits
- **Character Limit**: 25,000 characters per response
- **Request Timeout**: 30 seconds
- **Max Search Results**: 20 per query
- **Rate Limiting**: Respects GitHub API limits
## 🤝 Contributing
This MCP server is designed to be extensible. To add new features:
1. Add new tool methods with `@mcp.tool` decorator
2. Define Pydantic input models for validation
3. Follow the existing patterns for error handling
4. Add corresponding evaluation questions
## 📄 License
MIT License - Same as React Native Godot
## 🙏 Credits
- **React Native Godot**: Created by [Born](https://born.com) and [Migeran](https://migeran.com)
- **MCP Server**: Built using the Model Context Protocol by Anthropic
- **FastMCP**: Simplified MCP development framework
## 📞 Support
For React Native Godot questions:
- [GitHub Issues](https://github.com/borndotcom/react-native-godot/issues)
- [Migeran Support](https://migeran.com/contact) (commercial)
For MCP server issues:
- Open an issue in this repository
- Check the evaluation results for capability verification
## 🚦 Status
- ✅ All documentation sections accessible
- ✅ Intelligent search functionality
- ✅ Complete example code coverage
- ✅ Platform-specific guidance
- ✅ Comprehensive troubleshooting
- ✅ Direct file access from repository
- ✅ 10 evaluation questions for testing
---
Built with ❤️ for the React Native and Godot communities