Repository files navigation

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

• 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

If you're using Claude Code, simply run:

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

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

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

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

Or on macOS with Homebrew:

brew install uv

# Clone the repositorygit clone https://github.com/compiler-explorer/ce-mcpcd ce-mcp# Option A: Quick setup (Linux/macOS)./setup.sh# Option B: Manual setupuv venvsource .venv/bin/activate# On Windows: .venv\Scripts\activateuv pip install -e".[dev]"

# 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/

# Basic usagece-mcp# With custom configurationce-mcp --config~/.config/ce-mcp/config.yaml# With verbose loggingce-mcp --verbose

The server provides 12 specialized MCP tools for compilation, analysis, and sharing. Seedocs/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

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

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

# Install with development dependenciesuv pip install -e".[dev]"# Run tests.venv/bin/pytest tests/# Run integration tests (requires internet).venv/bin/pytest tests/ -m integration# Format codeblack.isort.# Type checkingmypy ce_mcp/# Code quality checksuv run black ce_mcp/ tests/isort ce_mcp/ tests/

If you prefer using pip instead of UV:

# Create virtual environmentpython3 -m venv .venvsource .venv/bin/activate# On Windows: .venv\Scripts\activate# Install dependenciespip install -e".[dev]"# Run the serverpython -m ce_mcp.cli

"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]"

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

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

BSD-2-Clause