MCP Cheat Engine Server
Provides safe, read-only access to memory analysis and debugging functionality through the Model Context Protocol, allowing users to examine computer memory for software development, security research, and educational purposes.
README
MCP Cheat Engine Server Documentation
📋 Table of Contents
- Overview
- Quick Start Guide
- Installation
- Configuration
- Using the Tools
- Safety & Security
- Troubleshooting
- Advanced Usage
- API Reference
- FAQ
🎯 Overview
The MCP Cheat Engine Server provides safe, structured access to memory analysis and debugging functionality through the Model Context Protocol (MCP). This tool is designed for:
- Software developers debugging applications
- Security researchers analyzing programs
- Students learning about computer memory and reverse engineering
- Game modders understanding game mechanics
⚠️ Important Safety Notice
This server operates in READ-ONLY mode for safety. It can read and analyze memory but cannot modify it. All operations are logged for security auditing.
🔧 Key Features
- ✅ Process enumeration and attachment
- ✅ Memory reading with multiple data types
- ✅ Pattern scanning and searching
- ✅ Assembly code disassembly
- ✅ Pointer chain resolution
- ✅ Cheat Engine table (.CT) import
- ✅ Safe Lua script analysis
- ✅ Comprehensive security controls
🚀 Quick Start Guide
Prerequisites
- Windows 10/11 (64-bit recommended)
- Python 3.9 or higher
- Administrator privileges (for memory access)
- Claude Desktop or compatible MCP client
30-Second Setup
- Download the server files to your computer
- Open PowerShell as Administrator
- Navigate to the server directory
- Install dependencies:
pip install -r requirements.txt - Start the server:
python server/main.py
First Use
- List processes: Use the
list_processestool to see available programs - Attach to a process: Use
attach_to_processwith a process ID - Read memory: Use
read_memory_regionto examine memory - Detach safely: Use
detach_from_processwhen done
🔌 Claude Desktop MCP Setup
📖 For detailed Claude Desktop setup instructions, see MCP_SETUP.md
Quick configuration summary:
- Find your Claude Desktop config:
%APPDATA%\Claude\claude_desktop_config.json(Windows) - Add the MCP server configuration:
{
"mcpServers": {
"cheat-engine": {
"command": "python",
"args": ["path\\to\\server\\main.py", "--debug", "--read-only"],
"cwd": "path\\to\\cheat-engine-server-python"
}
}
}
- Restart Claude Desktop
- Test: Ask Claude to "list processes using the MCP server"
📦 Installation
Step 1: System Requirements
- Operating System: Windows 10/11 (Primary), Linux/macOS (Limited)
- Python Version: 3.9, 3.10, 3.11, or 3.12
- Memory: 4GB RAM minimum, 8GB recommended
- Permissions: Administrator/root access required
Step 2: Download and Setup
# Clone or download the project
cd C:\your-desired-location
# Extract files if downloaded as ZIP
# Navigate to project directory
cd cheat-engine-server-python
# Verify Python version
python --version
Step 3: Install Dependencies
# Install required packages
pip install -r requirements.txt
# Verify installation
python -c "import mcp, trio, psutil, capstone; print('All dependencies installed successfully!')"
Step 4: Test Installation
# Test the server
python server/main.py --test
# You should see: "MCP Cheat Engine Server initialized successfully"
⚙️ Configuration
Basic Configuration
The server uses configuration files in the server/config/ directory:
settings.json (Auto-created on first run)
{
"security": {
"read_only_mode": true,
"require_whitelist": true,
"log_all_operations": true
},
"performance": {
"max_memory_read": 1048576,
"scan_timeout": 30,
"max_results": 1000
}
}
whitelist.json (Process Access Control)
{
"processes": [
{
"name": "notepad.exe",
"allowed": true,
"description": "Text editor for testing"
},
{
"name": "calculator.exe",
"allowed": true,
"description": "Calculator application"
}
]
}
Security Settings Explained
| Setting | Description | Default | Recommendation |
|---|---|---|---|
read_only_mode |
Prevents memory writing | true |
Keep enabled |
require_whitelist |
Only allow whitelisted processes | true |
Enable for safety |
log_all_operations |
Log every operation | true |
Enable for auditing |
max_memory_read |
Maximum bytes per read | 1MB | Adjust as needed |
🛠️ Using the Tools
1. Process Management
List Available Processes
# Find processes you can attach to
result = use_tool("list_processes")
What you'll see:
- Process name and ID
- Memory usage
- Whether it's accessible
- Security level required
Attach to a Process
# Attach to a specific process
result = use_tool("attach_to_process", {
"process_id": 1234
})
Best practices:
- Start with simple programs like Notepad
- Always detach when finished
- Check the whitelist if attachment fails
2. Memory Reading
Read Memory at Address
# Read 64 bytes starting at address 0x140000000
result = use_tool("read_memory_region", {
"address": "0x140000000",
"size": 64,
"data_type": "bytes"
})
Supported Data Types
bytes- Raw byte datastring- ASCII/UTF-8 textint32- 32-bit signed integeruint32- 32-bit unsigned integerint64- 64-bit signed integeruint64- 64-bit unsigned integerfloat- 32-bit floating pointdouble- 64-bit floating point
3. Memory Scanning
Search for Patterns
# Find all occurrences of a byte pattern
result = use_tool("scan_memory", {
"pattern": "48 8B 05 ?? ?? ?? ??", # ?? = wildcard
"start_address": "0x140000000",
"end_address": "0x141000000"
})
Pattern Format Examples
"41 42 43"- Find bytes 0x41, 0x42, 0x43"48 ?? 05 ?? ?? ?? ??"- Wildcards for unknown bytes"Hello World"- Search for ASCII text"00 00 00 01"- Find integer value 1
4. Code Analysis
Disassemble Assembly Code
# Disassemble 100 bytes of code
result = use_tool("disassemble_code", {
"address": "0x140001000",
"size": 100,
"architecture": "x64"
})
Analyze Data Structures
# Analyze memory for data structures
result = use_tool("analyze_structure", {
"address": "0x200000000",
"size": 256
})
5. Pointer Chains
Follow Multi-Level Pointers
# Resolve [[base + 0x10] + 0x20] + 0x30
result = use_tool("resolve_pointer_chain", {
"base_address": "0x140000000",
"offsets": [16, 32, 48] # 0x10, 0x20, 0x30 in decimal
})
🔒 Safety & Security
Read-Only Protection
The server cannot modify memory - it can only read and analyze. This prevents:
- Accidental program crashes
- Security vulnerabilities
- System instability
Process Whitelist
Only approved processes can be accessed:
{
"processes": [
{"name": "notepad.exe", "allowed": true},
{"name": "suspicious.exe", "allowed": false}
]
}
Operation Logging
All operations are logged to logs/operations.log:
2025-07-30 10:30:15 - INFO - Process attached: notepad.exe (PID: 1234)
2025-07-30 10:30:20 - INFO - Memory read: 0x140000000, size: 64
2025-07-30 10:30:25 - INFO - Process detached: notepad.exe
Permission Requirements
- Windows: Run as Administrator
- Linux: Run as root or with appropriate capabilities
- macOS: May require disabling SIP for some operations
🔧 Troubleshooting
Common Issues
"Access Denied" Error
Problem: Cannot attach to process Solutions:
- Run as Administrator
- Check if process is in whitelist
- Verify process is not protected by anti-virus
"Module Not Found" Error
Problem: Python dependencies missing Solution:
pip install --upgrade -r requirements.txt
"Process Not Found" Error
Problem: Process ID doesn't exist Solutions:
- Use
list_processesto get current IDs - Check if process is still running
- Try process name instead of ID
Memory Read Fails
Problem: Cannot read memory at address Solutions:
- Check if address is valid with
get_memory_regions - Verify memory protection allows reading
- Try smaller read size
Debug Mode
Enable detailed logging:
python server/main.py --debug
Getting Help
- Check the FAQ section below
- Review log files in
logs/directory - Verify configuration in
server/config/ - Test with simple programs like Notepad first
🎓 Advanced Usage
Cheat Engine Table Import
Import existing .CT files:
result = use_tool("import_cheat_table", {
"file_path": "C:/path/to/table.CT"
})
Lua Script Analysis
Analyze Cheat Engine Lua scripts:
result = use_tool("execute_lua_script", {
"script_content": "print('Hello from Lua')",
"safe_mode": true
})
Custom Memory Regions
Define specific regions for analysis:
# Get full memory map
regions = use_tool("get_memory_regions")
# Analyze specific region
for region in regions:
if region['protect'] == 'PAGE_EXECUTE_READ':
# Analyze executable memory
pass
Automation Examples
# Complete analysis workflow
def analyze_process(process_name):
# 1. Find and attach to process
processes = use_tool("list_processes")
target_pid = find_process_by_name(processes, process_name)
# 2. Attach to process
use_tool("attach_to_process", {"process_id": target_pid})
# 3. Get memory layout
regions = use_tool("get_memory_regions")
# 4. Scan for patterns
for region in regions:
if region['readable']:
scan_results = use_tool("scan_memory", {
"pattern": "48 8B 05",
"start_address": region['base_address'],
"end_address": region['base_address'] + region['size']
})
# 5. Clean up
use_tool("detach_from_process")
📚 API Reference
Tool Categories
Process Management
- list_processes() - Get all running processes
- attach_to_process(process_id) - Attach to specific process
- detach_from_process() - Safely detach from current process
- get_process_info() - Get detailed process information
Memory Operations
- read_memory_region(address, size, data_type) - Read memory
- get_memory_regions() - Get virtual memory layout
- scan_memory(pattern, start_address, end_address) - Pattern search
Analysis Tools
- analyze_structure(address, size) - Structure analysis
- disassemble_code(address, size, architecture) - Code disassembly
- resolve_pointer_chain(base_address, offsets) - Pointer resolution
Advanced Features
- import_cheat_table(file_path) - Import .CT files
- execute_lua_script(script_content, safe_mode) - Lua analysis
Data Types Reference
| Type | Size | Range | Use Case |
|---|---|---|---|
int8 |
1 byte | -128 to 127 | Small signed numbers |
uint8 |
1 byte | 0 to 255 | Bytes, characters |
int16 |
2 bytes | -32,768 to 32,767 | Short integers |
uint16 |
2 bytes | 0 to 65,535 | Port numbers |
int32 |
4 bytes | ±2.1 billion | Standard integers |
uint32 |
4 bytes | 0 to 4.2 billion | Addresses (32-bit) |
int64 |
8 bytes | ±9.2 quintillion | Large numbers |
uint64 |
8 bytes | 0 to 18.4 quintillion | Addresses (64-bit) |
float |
4 bytes | ±3.4E±38 | Decimal numbers |
double |
8 bytes | ±1.7E±308 | High precision decimals |
❓ FAQ
General Questions
Q: Is this tool safe to use? A: Yes, the server operates in read-only mode and cannot modify memory or harm your system.
Q: Can I use this on games? A: Yes, but respect the terms of service of online games. This tool is primarily for educational and debugging purposes.
Q: Do I need Cheat Engine installed? A: No, this is a standalone server that provides similar functionality through MCP.
Technical Questions
Q: Why do I need Administrator privileges? A: Windows requires elevated privileges to read memory from other processes for security reasons.
Q: Can I run this on Mac or Linux? A: The server has limited support for Mac/Linux. Some Windows-specific features may not work.
Q: How much memory does the server use? A: Typically 50-100MB, depending on the size of processes being analyzed.
Usage Questions
Q: What processes should I start with? A: Begin with simple programs like Notepad, Calculator, or your own test applications.
Q: How do I find the right memory addresses? A: Use memory scanning to find patterns, then analyze the results to identify relevant addresses.
Q: Can I save my analysis results? A: Yes, all tool results can be saved to files for later reference and analysis.
Troubleshooting Questions
Q: The server won't start - what should I check? A: Verify Python version (3.9+), install dependencies, and run as Administrator.
Q: I can't attach to a process - why? A: Check the process whitelist, verify the process is running, and ensure you have Administrator privileges.
Q: Memory reads are failing - what's wrong?
A: The memory address may be invalid or protected. Use get_memory_regions to find readable areas.
📝 License & Credits
This project is licensed under the MIT License. Built with:
- FastMCP - Model Context Protocol implementation
- Capstone - Disassembly engine
- psutil - Process and system utilities
- trio - Async I/O framework
📞 Support
For additional help:
- Review this documentation thoroughly
- Check the troubleshooting section
- Enable debug mode for detailed error information
- Test with simple programs first
Remember: This tool is for educational and legitimate debugging purposes. Always respect software licenses and terms of service.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。