Get the FREE Ultimate OpenClaw Setup Guide →

MDB

Multi Debugger MCP server that enables LLMs to interact with GDB and LLDB for binary debugging and analysis.

Installation
Run this command in your terminal to add the MCP server to Claude Code.
Run in terminal:
Command
claude mcp add --transport stdio smadi0x86-mdb-mcp uv run server.py

How to use

This MCP server implements a Multi-Debugger that can drive both LLDB and GDB debuggers through a single API surface. It exposes unified tools to list debuggers, start sessions, terminate sessions, and issue commands across different debuggers, making it suitable for integration with Claude Desktop, VSCode Copilot, or other AI-assisted environments. The server auto-detects the appropriate debugger (LLDB or GDB) and provides dedicated tooling paths for each debugger as well as a unified interface for common actions like starting a session, sending commands, and querying status. Use the provided functions to manage sessions without needing to manually switch between debuggers in your client.

Key capabilities include a unified toolbox (debugger_status, debugger_start, debugger_terminate, debugger_list_sessions, debugger_command) and specialized LLDB and GDB tooling (lldb_start, lldb_terminate, lldb_list_sessions, lldb_command; gdb_start, gdb_terminate, gdb_list_sessions, gdb_command). This enables you to programmatically control debugging sessions from your LLM client, automate workflows, and issue arbitrary debugger commands as needed.

How to install

Prerequisites: - Python (3.x) installed on your system - Access to the uv CLI tool used to run the MCP server (as shown in the Quick Start) — for example, via uv sync, uv venv, and uv run.

  1. Clone the repository or download the MCP server files to your environment.
  2. Create a Python virtual environment (optional but recommended):
    • uv venv (this sets up a virtual environment in the project context)
  3. Install or ensure the uv CLI is available in your PATH. If your environment requires it, install uv using your preferred method (as documented by the uv tool you use).
  4. Start the server:
    • uv run server.py
  5. Verify the server is running and reachable by exercising the available MCP endpoints or by following the usage examples in the README (e.g., starting sessions, listing debuggers).

Note: For macOS with experimental LLDB support, you may need to install LLVM (which includes LLDB) and Python bindings as described in the README:

  • brew install llvm python3
  • pip3 install mcp pygdbmi --break-system-packages

Additional notes

Notes and tips: - This MCP supports both GDB and LLDB; LLDB support is marked as experimental on macOS. The server will automatically choose the appropriate debugger when starting a session if not explicitly specified. - If you encounter issues with Python bindings for LLDB, ensure that you have LLVM and Python bindings installed as shown in the macOS installation section. - Use the unified debugger commands (debugger_command, debugger_start, debugger_terminate) for general operations, or use LLDB/GDB specific helpers (lldb_command, gdb_command) for advanced control. - You can verify availability of debuggers at runtime using the provided status checks (e.g., LLDB available, GDB available). - When using WSL or non-WSL environments, provide the full paths to Python and the server script as shown in the VSCode Copilot integration examples. - For testing, run the included tests via uv run python run-tests.py with appropriate flags (e.g., --check-deps, --type all).

Related MCP Servers

Sponsor this space

Reach thousands of developers