This MCP server provides a cross-platform interface for controlling debuggers such as WinDbg and LLDB, and for analyzing memory dumps. It exposes an MCP surface over HTTP (and a companion CLI) to manage sessions, dumps, analysis, and reporting. You can run the server in MCP mode to engage in interactive analysis, or enable the HTTP API to upload dumps, manage symbols, and fetch reports. The included dbg-mcp CLI lets you connect to the running server, upload dumps, run analyses (crash, .NET, perf, etc.), and generate formatted reports. The tools are designed to work together: create a session, open a dump, run crash/.NET analyses, and produce human- or machine-readable outputs. For AI-assisted analysis, you can enable the optional analyze ai workflow in the CLI when configured.

To use the HTTP API, first start the server with the MCP-over-HTTP mode, then access endpoints to upload dumps, manage symbols, and download reports. The server also ships with a compact MCP tool surface (11 tools) accessible through the CLI, including session management, dumps, analysis, comparison, reports, watches, symbol management, and raw debugger commands via exec. This makes it suitable for automated pipelines or interactive investigations alike.

Prerequisites

• Git
• .NET 10 SDK
• Optional: Docker if you prefer containerized runs (see ADVANCED.md for details)

Step-by-step installation

1. Clone the repository git clone https://github.com/tonyredondo/debugger-mcp-server.git cd debugger-mcp-server

2. Build the server (Release configuration is recommended) dotnet build -c Release

3. Run the server in MCP-over-HTTP mode dotnet run --project DebuggerMcp/DebuggerMcp.csproj -- --mcp-http

   The server will start and listen for MCP over HTTP (default behavior as described in the repo).

4. (Optional) Run the companion CLI to interact with the server dotnet run --project DebuggerMcp.Cli/DebuggerMcp.Cli.csproj

   In the dbg-mcp shell, connect to the server and start uploading/analyzing dumps.

5. Verify API health GET http://localhost:5000/health

Notes

• If you prefer Docker, see ADVANCED.md in the repository for multi-platform images and environment configurations.
• The exact port and endpoints may vary if you customize startup flags; refer to ADVANCED.md for full configuration options.

Tips and considerations:

• Ensure you have the required debugger tools installed on the host (WinDbg on Windows, LLDB on Linux/macOS) if you plan to perform debugging or dump analysis locally.
• The HTTP API endpoints support dump management, symbol management, and report generation. Use the CLI for quick interactive workflows or integrate the HTTP API into automation scripts.
• If you enable AI analysis in the CLI (analyze ai), you must configure MCP sampling and an LLM provider as described in the repository docs.
• Environment variables (as highlighted in ADVANCED.md) control feature flags such as symbol enrichment, data sources, and performance tracing. Common knobs include enabling/disabling GitHub enrichment, heap enumeration safety valves, and symbol workflows.
• When operating in production, consider running behind a reverse proxy and enabling TLS, as well as proper authentication/authorization for the HTTP API endpoints.