SetsunaYukiOvO/x64dbg-mcp

What it solves

This project provides a bridge between the x64dbg/x32dbg debuggers and AI agents. It allows external applications and AI models to programmatically control a debugger, perform reverse engineering tasks, and analyze binary execution in real-time via a standardized protocol.

How it works

The project is a plugin for x64dbg and x32dbg that implements the Model Context Protocol (MCP). It starts an HTTP server (defaulting to port 3000) that exposes a JSON-RPC 2.0 interface. This allows AI agents to interact with the debugger through three core components:

• Tools: 79 AI-invokable functions for execution control (run/pause/step), memory and register access, breakpoint management, and disassembly.
• Resources: 15 context providers that give the AI access to debugger state, registers, modules, and memory maps.
• Prompts: 10 workflow templates for common tasks like crash analysis, vulnerability hunting, and binary unpacking.

Who it’s for

Reverse engineers, security researchers, and developers who want to use AI agents (like Cursor or other MCP-compatible clients) to automate and assist in the debugging and analysis of x86/x64 binaries.

Highlights

• Comprehensive Control: Full access to execution, memory, registers, and thread management.
• Dual Architecture Support: Works with both x64 and x86 architectures.
• Harnesses MCP: Fully compliant with the Model Context Protocol for easy integration with AI tools.
• Security-First: Includes permission-based access control and optional Bearer token authentication to prevent unauthorized debugger control.
• Integrated Workflow Templates: Pre-defined prompts for complex tasks like OEP detection and algorithm reversing.