miscusi-peek/cheatengine-mcp-bridge

Connect Cursor, Copilot & Claude AI directly to Cheat Engine via MCP. Automate reverse engineering, pointer scanning, and memory analysis using natural language.

What it solves

Manual memory analysis, reverse engineering, and game modding typically require days of tedious work scanning hex dumps and tracing pointer chains. This project eliminates that manual effort by allowing users to interact with a program's memory using natural language via an AI agent.

How it works

It implements a bridge between an AI agent (like Claude, Cursor, or Copilot) and Cheat Engine using the Model Context Protocol (MCP).

  1. AI Agent: Sends requests via JSON-RPC over stdio.
  2. MCP Server: A Python server (mcp_cheatengine.py) that translates AI requests into commands.
  3. Communication: The server communicates with Cheat Engine via Windows Named Pipes (or a TCP relay for remote/WSL access).
  4. Cheat Engine: A Lua script (ce_mcp_bridge.lua) runs inside Cheat Engine to execute the memory operations and return results to the AI.

Who it’s for

Reverse engineers, security auditors, modders, and developers who want to accelerate the process of analyzing program memory, finding specific opcodes, or creating game bots and trainers.

Highlights

  • Extensive Toolset: Provides ~180 MCP tools covering memory reading/writing, pointer chain resolution, AOB scanning, and code disassembly.
  • Advanced Analysis: Supports C++ object identification via RTTI and automatic structure dissection.
  • Deep Debugging: Integrates hardware breakpoints and Ring -1 invisible tracing via DBVM.
  • Process Control: Ability to open, create, pause, and resume target processes.
  • Code Injection: Supports DLL injection and compiling C code into injected shellcode.
  • TCP Relay: Allows the MCP server to run on a different machine or environment (like Linux/WSL) while Cheat Engine runs on Windows.