AI Briefing
KO

Pickx64dbg-mcp-server: An MCP Server for LLM Agents to Directly Control x64dbg

duty1g/x64dbg-mcp-server

·2026.08.24 20:05

Perform reverse engineering using only natural language instructions, without the need to manually manipulate the debugger. This MCP server, which installs natively into x64dbg, exposes the debugger's full functionality via an HTTP API, allowing MCP-compatible AI assistants to programmatically control everything from setting breakpoints to reading memory and dumping registers.

Written in Zig, it has no runtime or external dependencies and supports both x32 and x64 architectures as a single binary. It requires no .NET or Python environment; simply copy it to the plugin folder for immediate operation, and you can cross-compile Windows plugins on Linux or macOS.

It provides 84 MCP tools and 22 event callbacks, covering professional debugging workflows such as disassembly, pattern scanning, PE analysis, and OEP detection. It supports dual transport methods—Streamable HTTP and SSE—for compatibility with both modern and legacy MCP clients, and enforces Bearer token authentication on all requests to ensure security.

Ideal for developers looking to automate repetitive debugging tasks or integrate AI analysis capabilities into their debugging environment. Unlike traditional methods that require manually entering complex commands, the AI understands context and selects and executes the necessary tools, reducing the cognitive load of the reverse engineering process.

GitHub
GitHub repository

duty1g/x64dbg-mcp-server

x64dbg-MCP Server is a native MCP (Model Context Protocol) plugin for x64dbg that exposes the debugger's full functionality over HTTP. Connect any MCP-compatible AI assistant and control x64dbg programmatically: set breakpoints, step through code, read memory, dump registers, and more. Built with Zig — zero dependencies, single-binary output, cros

Zig

This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.