
x64dbg 调试器的 MCP 服务器插件——使 AI 代理和外部工具能够通过 HTTP/SSE 的 JSON-RPC 2.0 控制调试
概览
这是适用于x64dbg和x32dbg的模型上下文协议(MCP)服务器实现,支持通过JSON-RPC 2.0接口进行远程调试。该插件允许外部应用程序和AI代理以程序方式与调试器交互。 - : 实现所有三个核心MCP构建模块 - : AI调用调试功能 - : 应用控制上下文数据源 - : 用户引导调试工作流模板 - : 标准且语言无关的接口 - (MCP 2025-03-26)在/mcp上,以及为旧客户端在/sse上提供遗留HTTP+SSE接口 - : - 执行控制(初始化/运行/暂停/步run_to/重启/停止) - 内存读写/搜索/分配 - 寄存器访问(50+个寄存器,包括GPR、SSE、AVX)- 断点管理(软件、硬件、内存、条件、日志)- 反汇编和符号解析 - 线程管理(列表、切换、暂停、恢复)- 栈跟踪和分析(模块转储、内存转储、封包器检测、OEP检测)- (执行x64dbg命令、批处理操作)-...
概述(AI 翻译)
英文 |中文
这是适用于x64dbg和x32dbg的模型上下文协议(MCP)服务器实现,支持通过JSON-RPC 2.0接口进行远程调试。该插件允许外部应用程序和AI代理以程序方式与调试器交互。
现在支持x64和x86两种架构!
特色
-
完全MCP规范合规:实现所有三个核心MCP构建模块
-
工具(79):AI调用调试功能
-
资源(7 + 8个模板):应用程序控制的上下文数据源
-
提示词(10):用户引导调试工作流程模板
-
JSON-RPC 2.0 协议:标准、语言无关接口
-
可流式HTTP传输(MCP 2025-03-26)在
/mcp上,加上/sse上的遗留HTTP+SSE,用于较老的客户端 -
工具 - AI控制调试(79个功能):
-
执行控制(init/run/pause/step/run_to/restart/stop)
-
内存读/写/搜索/分配
-
寄存器访问(50+寄存器,包括GPR、SSE、AVX)
-
断点管理(软件、硬件、内存、条件、日志)
-
反汇编和符号解析
-
线程管理(列表、交换、暂停、恢复)
-
栈迹与分析
本页由 AI 翻译,技术细节请以原文为准。
README
以下为原文 README 节选,完整内容请查看原文。
x64dbg MCP Server Plugin
English | 中文
A Model Context Protocol (MCP) server implementation for x64dbg and x32dbg, enabling remote debugging through a JSON-RPC 2.0 interface. This plugin allows external applications and AI agents to interact with the debugger programmatically.
Now supports both x64 and x86 architectures!
Features
-
Full MCP Specification Compliance: Implements all three core MCP building blocks
- Tools (79): AI-invokable debugging functions
- Resources (7 + 8 templates): Application-controlled context data sources
- Prompts (10): User-guided debugging workflow templates
-
JSON-RPC 2.0 Protocol: Standard, language-agnostic interface
-
Streamable HTTP transport (MCP 2025-03-26) on
/mcp, plus legacy HTTP+SSE on/ssefor older clients -
Tools - AI-Controlled Debugging (79 functions):
- Execution control (init/run/pause/step/run_to/restart/stop)
- Memory read/write/search/allocate
- Register access (50+ registers including GPR, SSE, AVX)
- Breakpoint management (software, hardware, memory, conditional, logging)
- Disassembly and symbol resolution
- Thread management (list, switch, suspend, resume)
- Stack trace and analysis
- Dump & Analysis (module dump, memory dump, packer detection, OEP detection)
- Script execution (execute x64dbg/x32dbg commands, batch operations)
- Context snapshots (capture and compare debugging state)
-
Resources - Context Providers (7 direct + 8 templates):
- Direct resources: debugger state, registers, modules, threads, memory map, breakpoints, stack
- Resource templates: memory content, disassembly, module info, symbol resolution, function analysis
- Read-only, application-controlled access
-
Prompts - Workflow Templates (10 prompts):
- Crash analysis, vulnerability hunting, function tracing
- Binary unpacking, algorithm reversing, execution comparison
- String hunting, code patching, API monitoring
- Debug session initialization
-
Security: Permission-based access control
-
Extensible: Plugin architecture for custom methods, resources, and prompts
Prerequisites
- Windows 10/11 (64-bit host; supports building both x64 and x86 plugins)
- CMake 3.15 or higher
- Visual Studio 2022 with C++ Desktop Development workload
- vcpkg - Package manager for C++ libraries
- Git - For cloning the repository
Quick Build
The easiest way to build is using the provided build script:
# Clone the repository
git clone https://github.com/SetsunaYukiOvO/x64dbg-mcp.git
cd x64dbg-mcp
# Build both x64 and x86 architectures (recommended)
.\build.bat
# Build only x64 architecture
.\build.bat --x64-only
# Build only x86 architecture
.\build.bat --x86-only
# Clean rebuild
.\build.bat --clean
# The script will:
# 1. Automatically detect vcpkg installation
# 2. Download dependencies (nlohmann_json)
# 3. Configure CMake for both architectures
安装
This server does not publish a one-line install command.
Open the repository installation guide 新手?先看 mcpskill.org 教程 →配置
{
"mcpServers": {
"x64dbg": {
"type": "http",
"url": "http://127.0.0.1:3000/mcp"
}
}
}