DuinoDu/install-x

install-x

A comprehensive MCP (Model Context Protocol) server that automates library installation across multiple programming languages including Python, JavaScript, Rust, C, and C++. This server provides intelligent library detection and installation capabilities with support for complex dependencies like CUDA, PyTorch, and specialized ML libraries.

Features

• Multi-language Support: Python, JavaScript/Node.js, Rust, C, and C++
• Smart Detection: Automatically detects available package managers and libraries
• CUDA Integration: Intelligent CUDA version detection for PyTorch and ML libraries
• Comprehensive Library Coverage: Supports popular libraries across all languages
• Cache Management: Efficient caching system to avoid redundant installations
• MCP Server: Full Model Context Protocol integration for AI assistants

Installation

Prerequisites

• Python 3.8 or higher
• Node.js and npm (for JavaScript packages)
• Cargo (for Rust packages)
• CUDA (optional, for ML libraries)

Setup

1. Clone the repository:

git clone https://github.com/DuinoDu/install-x
cd install-x

2. Install dependencies using rye:

rye sync

3. Activate the virtual environment:

source .venv/bin/activate

Usage

As MCP Server

Using MCP Inspector

npx @wong2/mcp-cli python3 src/install_x/mcp_server.py

Using Claude Desktop

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "install-x": {
      "command": "python3",
      "args": ["/path/to/install-x/src/install_x/mcp_server.py"]
    }
  }
}

Available Tools

1. is_supported(library_name)

Check if a library is supported for installation.

Parameters:

• library_name (str): Name of the library to check

Returns:

• library_name: The normalized library name
• is_supported: Boolean indicating if the library is supported

Example:

result = is_supported("numpy")
# Returns: {"library_name": "numpy", "is_supported": true}

2. install(library_name)

Install a library in the current directory.

Parameters:

• library_name (str): Name of the library to install

Returns:

• success: Boolean indicating installation success
• message: Human-readable status message
• details: Detailed installation results for each language

Example:

result = install("torch")
# Returns detailed installation results

Supported Libraries

Python

• ML/AI: torch, MinkowskiEngine, spconv, finegrasp_graspnet1b, graspnetAPI
• General: numpy, pandas, matplotlib, requests, flask, django
• Data Science: scikit-learn, tensorflow, opencv-python, pillow

JavaScript/Node.js

• Frameworks: express, react, vue, angular, next.js
• Utilities: lodash, axios, moment, chalk
• Development: typescript, webpack, jest, eslint, prettier

Rust

• Web: tokio, axum, rocket, reqwest
• CLI: clap, anyhow, thiserror
• Database: diesel, sqlx
• Serialization: serde

C

• System: ssl, crypto, curl, zlib, sqlite3
• Graphics: png, jpeg, opengl
• XML: libxml2, libxslt

C++

• Boost: boost-all-dev
• Computer Vision: opencv-dev
• Math: eigen3-dev
• ML: tensorflow-dev
• 3D: pcl-dev

Installation Scripts

The project includes specialized installation scripts for complex libraries:

Available Scripts

• torch.sh: PyTorch with CUDA support and version detection
• MinkowskiEngine.sh: NVIDIA Minkowski Engine for sparse tensor operations
• spconv.sh: Sparse convolution library for 3D data
• finegrasp_graspnet1b.sh: Grasp detection models
• graspnetAPI.sh: GraspNet API installation
• RoboTwin.sh: Robotics simulation tools

Script Features

• Automatic CUDA Detection: Detects CUDA version and selects appropriate PyTorch builds
• Cache Management: Uses local cache to speed up repeated installations
• Dependency Resolution: Handles complex dependency chains
• Error Handling: Provides helpful error messages and troubleshooting guides

Testing

Run the comprehensive test suite:

python -m pytest tests/

Or run specific tests:

# Test library support detection
python -m pytest tests/test_mcp_server.py::TestLibraryInstaller::test_is_supported_known_python_library

# Test numpy installation (as requested in PRD)
python -m pytest tests/test_mcp_server.py::TestLibraryInstaller::test_numpy_specific_installation

# Run all tests with verbose output
python -m pytest tests/ -v

Development

Project Structure

install-x/
├── src/install_x/
│   ├── __init__.py
│   ├── mcp_server.py          # Main MCP server implementation
│   └── install/               # Installation scripts
│       ├── base.sh            # Common utilities
│       ├── torch.sh           # PyTorch installation
│       ├── MinkowskiEngine.sh # NVIDIA Minkowski Engine
│       ├── spconv.sh          # Sparse convolution
│       ├── _template.sh       # Template for new scripts
│       └── *.patch            # Patches for specific libraries
├── tests/
│   └── test_mcp_server.py     # Comprehensive test suite
├── pyproject.toml             # Project configuration
└── requirements.lock          # Locked dependencies

Adding New Libraries

1. For simple libraries: Add to the appropriate language category in LibraryInstaller.SUPPORTED_LANGUAGES
2. For complex libraries: Create a new installation script in src/install_x/install/
3. For patches: Create .patch files alongside installation scripts

Environment Variables

• INSTALL_X_CACHE: Custom cache directory (default: ~/.cache/install-x)
• FORCE_CUDA_VERSION: Force specific CUDA version (e.g., "cu121", "cu124")
• FORCE_INSTALL_TORCH: Force PyTorch reinstallation even if already installed

Examples

Installing PyTorch with CUDA Support

# The server will automatically detect CUDA version
python3 src/install_x/mcp_server.py
# Then call: install("torch")

Installing Multiple Libraries

# Check support for various libraries
python3 -c "
from install_x.mcp_server import is_supported
libraries = ['numpy', 'react', 'tokio', 'opencv-dev']
for lib in libraries:
    print(f'{lib}: {is_supported(lib)}')
"

Integration with AI Assistants

When used as an MCP server, AI assistants can:

1. Query library availability before installation
2. Install required dependencies automatically
3. Handle complex installation scenarios (CUDA, GPU support, etc.)
4. Provide installation feedback and troubleshooting

Troubleshooting

Common Issues

1. CUDA Detection Issues

   • Ensure CUDA is properly installed: nvcc --version
   • Set FORCE_CUDA_VERSION if auto-detection fails

2. Permission Errors

   • Run with appropriate permissions for system-wide installations
   • Use virtual environments when possible

3. Network Issues

   • Check internet connectivity for package downloads
   • Configure proxy settings if behind corporate firewall

4. Cache Issues

   • Clear cache: rm -rf ~/.cache/install-x
   • Reset virtual environments if needed

Getting Help

• Check the test suite for usage examples
• Review installation scripts for specific library requirements
• Examine error messages in the detailed installation results

License

This project is provided as-is for educational and development purposes. Individual libraries installed may have their own licenses.