Compiler Explorer MCP Server

A Model Context Protocol (MCP) server that provides efficient access to Compiler Explorer (godbolt.org) compilation and execution services.

Features

• Token-efficient responses - Minimized output while preserving essential information
• Use-case driven tools - Specific functions for common compilation scenarios
• Smart filtering - Context-aware output filtering based on the task
• Comprehensive language support - C++, C, Rust, Go, Python, and more
• Claude Code integration - Automated setup for Claude Code workflows

Quick Start (Claude Code Users)

If you're using Claude Code, simply run:

git clone https://github.com/compiler-explorer/ce-mcp
cd ce-mcp
./setup-claude-code.sh

This will automatically install and configure everything. That's it!

Quick Setup

Prerequisites

• Python 3.10+ - Required for running the MCP server
• UV - Fast Python package manager (recommended)

1. Install UV (if not already installed)

curl -LsSf https://astral.sh/uv/install.sh | sh

Or on macOS with Homebrew:

brew install uv

2. Clone and Setup

# Clone the repository
git clone https://github.com/compiler-explorer/ce-mcp
cd ce-mcp

# Option A: Quick setup (Linux/macOS)
./setup.sh

# Option B: Manual setup
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e ".[dev]"

3. Run Tests (Optional)

# Run unit tests
.venv/bin/pytest tests/ -m "not integration"

# Run integration tests (requires internet)
.venv/bin/pytest tests/ -m integration

# Run all tests
.venv/bin/pytest tests/

4. Start the MCP Server

# Basic usage
ce-mcp

# With custom configuration
ce-mcp --config ~/.config/ce-mcp/config.yaml

# With verbose logging
ce-mcp --verbose

Available Tools

The server provides 12 specialized MCP tools for compilation, analysis, and sharing. See docs/available_tools.md for detailed documentation of each tool.

Core Tools:

• compile_check_tool - Syntax validation
• compile_and_run_tool - Compile and execute
• compile_with_diagnostics_tool - Error analysis

Analysis Tools:

• analyze_optimization_tool - Assembly optimization analysis
• compare_compilers_tool - Compiler comparison
• lookup_instruction_tool - Assembly instruction documentation

Discovery & Sharing:

• generate_share_url_tool - Create Compiler Explorer URLs
• find_compilers_tool - Find compilers with filtering
• get_libraries_tool - List available libraries
• get_library_details_tool - Library information
• get_languages_tool - List supported languages
• download_shortlink_tool - Download shared code

Configuration

Create a configuration file at ~/.config/compiler_explorer_mcp/config.yaml:

compiler_explorer_mcp:
  api:
    endpoint: "https://godbolt.org/api"
    timeout: 30
    
  defaults:
    language: "c++"
    compiler: "g132"
    
  output_limits:
    max_stdout_lines: 100
    max_stderr_lines: 50

Development

Setting up for Development

# Install with development dependencies
uv pip install -e ".[dev]"

# Run tests
.venv/bin/pytest tests/

# Run integration tests (requires internet)
.venv/bin/pytest tests/ -m integration

# Format code
black .
isort .

# Type checking
mypy ce_mcp/

# Code quality checks
uv run black ce_mcp/ tests/
isort ce_mcp/ tests/

Alternative: Traditional pip setup

If you prefer using pip instead of UV:

# Create virtual environment
python3 -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -e ".[dev]"

# Run the server
python -m ce_mcp.cli

Troubleshooting

Common Issues

"command not found: ce-mcp"

• Make sure your virtual environment is activated
• Try running directly: python -m ce_mcp.cli

"ModuleNotFoundError: No module named 'packaging'"

• Install missing dependency: uv pip install packaging
• Or reinstall with all dependencies: uv pip install -e .

"ModuleNotFoundError: No module named 'ce_mcp'"

• Install in development mode: uv pip install -e .
• Check your virtual environment is activated

Integration tests failing

• Tests require internet access to godbolt.org
• Some tests may be skipped if the API is temporarily unavailable
• Run unit tests only: pytest tests/ -m "not integration"

"uv not found"

• Install UV: curl -LsSf https://astral.sh/uv/install.sh | sh
• Or use pip: pip install -e ".[dev]"

Configuration

Create ~/.config/ce-mcp/config.yaml for custom settings:

compiler_explorer_mcp:
  api:
    timeout: 30
  defaults:
    language: "c++"
    compiler: "g132"
  output_limits:
    max_stdout_lines: 50

License

BSD-2-Clause