Skip to content
KitploitKITPLOIT
ToolsExploitsBlog
Log in
Submit
ToolsExploitsBlog
Submit

Hacking, PenTest, and Cybersecurity Tools for Your Security Arsenal!

Kitploit is a directory of hacking, cybersecurity, and pentesting tools. Discover the latest project updates to find vulnerabilities, analyze systems, automate testing, and strengthen your security.

FeedsContactPrivacy© 2026 Kitploit

Tool Directory

Categories

View all categories
Loading categories
MCPDebugger — A MCP Debugger Server for Windows executables (x86 and x64). Exposes debugger functionality as MCP Tools for static / dynamic analysis of the debuggee and memory inspection through a clean interface. | Kitploit
Tools/GitHubGitHub/fatmike-gh/mcpdebugger
Static AnalysisDynamic Analysis (Sandboxing)Memory ForensicsReverse EngineeringDebuggersBinary AnalysisAI-Assisted Reversing
GitHubfatmike-gh/mcpdebugger

MCPDebugger

A MCP Debugger Server for Windows executables (x86 and x64). Exposes debugger functionality as MCP Tools for static / dynamic analysis of the debuggee and memory inspection through a clean interface.

View Repository
1311043h 56m agoNot yet reviewed
Share

Fatmike's MCP Debugger Server

Introduction

This project is a self-made, lightweight MCP Debugger Server developed for learning and experimentation. The MCP Debugger Server exposes debugger functionality as MCP tools, enabling external MCP clients to control execution, inspect memory, and analyze the debuggee through a clean interface.

The debugger is written in c++ and implements core debugging functionality from scratch without relying on existing debugger frameworks. For instruction decoding the Zydis disassembler is used. The debugger is integrated with an MCP server written in c#, built using the official MCP C# SDK.

Keywords:
MCP server, debugger, disassembler, static analysis, dynamic analysis, debugging, AI, EXE, executable, Windows

Demo

image

Gemini CLI was used as MCP Client in the Demo

Quickstart: Setting up MCP.Server.exe in VS Code

This guide provides a step-by-step process for building and integrating the MCP.Server.exe into Visual Studio Code on Windows.

Step 1: Build the Solution

Before configuring VS Code, you must compile the mcp server:

  1. Open MCPDebugger.slnx in Visual Studio 2026. Required workloads: Desktop development with C++ including the components C++/CLI support for v145 build tools and vcpkg package manager, and .NET desktop development (it installs the .NET 10 SDK the solution targets). Older Visual Studio versions do not have the v145 toolset and cannot build the solution.
  2. Build the solution
  3. Confirm that MCP.Server.exe exists in your \build\Release\x86 or \build\Release\x64 subfolder.

The first build of a configuration and platform needs access to GitHub: the native test project gets Google Test through vcpkg, and vcpkg reads its package registry from there whenever it installs into a fresh build folder. Later builds of the same configuration and platform work offline. vcpkg sends usage data to Microsoft unless the environment variable VCPKG_DISABLE_METRICS is set.

MCP.Server.exe needs the .NET 10 runtime in its own architecture: the x86 server needs the x86 runtime, the x64 server the x64 runtime.

Step 2: Install Required VS Code Extensions

Ensure the following extensions are installed and active in VS Code:

  • GitHub Copilot
  • GitHub Copilot Chat

Step 3: Add the MCP Server

  1. Press Ctrl + Shift + P to open the Command Palette.
  2. Type > and select "Add MCP Server".
  3. Choose command (stdio) as the server type.

Step 4: Configure the Path

Enter the full path to MCP.Server.exe you built in Step 1.

[!CAUTION] Only one architecture (x86 or x64) should be active as MCP server at a time. If you switch architecture, remove the previous server entry first.

ArchitectureExample Path to Enter
Windows x64C:\YourProject\bin\x64\MCP.Server.exe
Windows x86C:\YourProject\bin\x86\MCP.Server.exe

Step 5: Finalize & Use

  1. Name the Server: Enter a name for the MCP Server and press Enter.
  2. Verify: Open Copilot Chat (Ctrl + Alt + I), switch to Agent Mode, and click the Tools icon. Your MCP tools should now be listed and ready for use. MCP tools can be enabled / disabled here.

VS Code Screenshots

VS Code: Toolbar with installed extensions and MCP Server

image

VS Code: Copilot Chat Window

image

Model Recommendations

For complex debugging and reverse-engineering tasks using this MCP debugger, Claude Opus 4.5 delivered by far the best results as of November 2025. It consistently handled large execution traces, detailed disassembly, and multi-step reasoning more reliably than other tested models.

ChatGPT 4.1 and Gemini 2.5 Pro were sufficient for simpler analysis scenarios, such as basic stepping, instruction inspection, and straightforward control-flow reasoning. These models performed well for the examples provided in the Example_x64 and Example_x86 folders, but showed limitations when dealing with more complex or long-running debugging sessions.

Model performance may change over time, but the above reflects observed behavior at the time of testing.

Disclaimer

Debugging via MCP can result in high token usage and increased costs. AI interactions are stateless, which means that all relevant debugging context (registers, memory, disassembly, modules, etc.) must be transferred repeatedly with each request.

Operations that frequently stop execution and return detailed state—such as stepping and temporary breakpoints—can significantly amplify token consumption.

If low-cost operation is required, it is strongly recommended to focus on static analysis and disable the following MCP tools:

  • Debugger_ContinueToBreakpoint
  • Debugger_ContinueToApiBreakpoint
  • Debugger_StepInto
  • Debugger_StepOver

Disabling these tools reduces the amount of repeated state transfer and helps keep token usage under control.

Examples

The repository includes example projects to help get started with static analysis and debugging and are located in the following folders:

  • Example_x64 and Example_x86
    The Example_x64 and Example_x86 projects provide a simple console application used as an introductory debugging target. The application prompts the user to enter a password at runtime. The correct password for both examples is secret.

  • MCP_Debugging_Sessions
    A folder containing more complex debugging sessions that leverage dynamic analysis and debugging. In these sessions Claude Opus 4.5 was used. The targets can be downloaded here. Please note that the targets were unpacked using UPX -d before using the MCP Debugger.

MCP Tools

The MCP server exposes the following tools to interact with the debugger:

Debugger Lifecycle

  • Debugger_Start
    Starts the debugger for the specified executable and breaks on the initial system breakpoint.

  • Debugger_Stop
    Stops the debugger and terminates the debuggee process.

Execution Control

  • Debugger_Continue
    Continues execution when the debuggee is stopped at a breakpoint.

  • Debugger_ContinueToEntryPoint
    Continues execution from the system breakpoint to the program entry point.

  • Debugger_ContinueToBreakpoint
    Continues execution until a temporary breakpoint at a specified virtual address is reached.

  • Debugger_ContinueToApiBreakpoint
    Continues execution until a temporary breakpoint on a specified API function is reached.

  • Debugger_StepInto
    Steps into the next instruction, entering function calls.

  • Debugger_StepOver
    Steps over function calls and continues to the next instruction.

Code & Memory Inspection

  • Debugger_Disassemble
    Disassembles memory at a given virtual address and returns instructions including symbol information.

  • Debugger_TryResolveString
    Attempts to resolve and read a string from the debuggee’s memory.

  • Debugger_ReadMemory
    Reads raw memory bytes from the debuggee process.

  • Debugger_WriteMemory
    Writes arbitrary bytes into the debuggee’s memory.

Module Inspection

  • Debugger_GetModules
    Returns a list of currently loaded modules for the debuggee process.

  • Debugger_GetModuleImports
    Returns the imported functions of a specified module.

Download Tool

Most Popular

View all →

Discover the most used tools by our community.

  • 1

    OsintHunter

    letox4/osinthunter

    🎯 The ultimate OSINT toolkit: investigation graph, confidence scoring, alias families, map, timeline and reports. 49 modules, 400+ sites, no API keys required.

    867
  • 2

    ghostlock-app

    yukonga/ghostlock-app

    GhostLock One-Tap Execution App (CVE-2026-43499)

    544
  • 3

    ARTEX

    baarai/artex

    LLM multi-agent autonomous penetration testing platform with planner/worker agents, asset and exploration graphs, MITM traffic recording, and vulnerability retesting.

    385
  • 4

    redforge

    cybersecurity91/redforge

    Browser-side red team toolkit with 102 offline tools for recon, payload generation, credential attacks, crypto, encoding and engagement reporting.

    360
  • 5

    looking-glass

    anonghosty/looking-glass

    CSIRT platform that ingests Shadowserver reports into MongoDB and replays them as interactive maps, analysis, findings, advisories and executive dashboards.

    302
  • 6

    CyberStrikeAI

    ed1s0nz/cyberstrikeai

    The system of action for AI-native cybersecurity—where intent becomes governed execution, evidence becomes operational memory, and every operation improves the next.

    279

Explore all tools

Browse our collection of tools

View all tools →