Cheat Engine

🎮 cheat-engine-mcp

AI-powered game memory scanner and reverse engineering toolkit — an MCP (Model Context Protocol) server that gives AI coding assistants direct access to memory scanning, process memory editing, and IL2CPP reverse engineering tools across Linux & Windows.

Turn your AI assistant into a game hacking partner. Scan memory, freeze values, search IL2CPP metadata, and modify game values — all through natural language.

✨ Features

78 MCP tools in total — all accessible via natural language through any MCP-compatible AI assistant.

🚀 Quick Start

1. Install Dependencies

Linux

# Arch Linux
sudo pacman -S scanmem gdb rust

# Debian / Ubuntu
sudo apt update && sudo apt install -y scanmem gdb cargo

Windows

No external dependencies (scanmem or GDB) required! The memory engine runs natively via Win32 API (kernel32.dll).

Install Rust if building from source:

winget install Rustlang.Rustup

2. Install on Linux

git clone https://github.com/gede-cahya/cheat-engine-mcp.git
cd cheat-engine-mcp
./install.sh

This builds the release binary and installs it to:

~/.local/bin/cheat-engine-mcp

Add this MCP config to your client:

{
  "mcpServers": {
    "cheat-engine-mcp": {
      "command": "/home/YOUR_USER/.local/bin/cheat-engine-mcp"
    }
  }
}

3. Install on Windows

Build the MCP server natively (or run install-antigravity.ps1):

git clone https://github.com/gede-cahya/cheat-engine-mcp.git
cd cheat-engine-mcp
cargo build --release

Copy the binary to a stable location:

New-Item -ItemType Directory -Force C:\Tools | Out-Null
Copy-Item .\target\release\cheat-engine-mcp.exe C:\Tools\cheat-engine-mcp.exe

Check it:

C:\Tools\cheat-engine-mcp.exe --help

Add this MCP config to your client:

{
  "mcpServers": {
    "cheat-engine-mcp": {
      "command": "C:\\Tools\\cheat-engine-mcp.exe"
    }
  }
}

Or run the automated setup script for Antigravity:

.\install-antigravity.ps1

4. Connect to Your AI Assistant

Choose your AI assistant and follow Install cheat-engine-mcp ke MCP Client.

Need an agent to install it for you? Copy docs/SKILL_INSTALL_PROMPT.md into Antigravity, Claude Code, or another coding agent.

🔧 MCP Client Configuration

All MCP-compatible clients use the same server binary. The only difference is where the config file lives.

Config File Locations

🧠 Skill / Plugin Installation

For the best experience, install the skill definition so your AI assistant knows how to use the tools effectively.

Google Antigravity

mkdir -p ~/.gemini/config/skills/cheat-engine-mcp
cp .claude/skills/cheat-engine-mcp/SKILL.md ~/.gemini/config/skills/cheat-engine-mcp/SKILL.md

Claude Code

mkdir -p YOUR_REPO/.claude/skills
cp -r .claude/skills/cheat-engine-mcp/ YOUR_REPO/.claude/skills/cheat-engine-mcp/
cp CLAUDE.md YOUR_REPO/CLAUDE.md

🎯 Usage Examples

Once configured, just talk to your AI assistant naturally:

Find and modify a game value

"Scan game TaskbarHero for health value 100, then set it to 999"

The AI will:

1. process_search → find the game PID
2. session_create → create a scan session
3. scanmem_scan_exact → scan for value 100
4. Refine with scanmem_scan_decreased / scanmem_scan_unchanged
5. scanmem_preview_write → preview the write
6. scanmem_write_selected → apply with safety guards

Reverse engineer a Unity/IL2CPP game

"Search for Hero class methods in TaskbarHero's IL2CPP dump"

The AI will:

1. workspace_set_active → set game workspace
2. il2cpp_class_search → find Hero class
3. il2cpp_method_search → find relevant methods
4. il2cpp_field_search → find class fields and offsets
5. il2cpp_find_by_rva → map RVA addresses

Automate IL2CPP Dump & Offset Recovery After Game Update

"Game Taskbar Heroes habis update, tolong dump dan scan offset terbarunya untuk godmode dan damage"

The AI will:

1. process_search → find TaskbarHero.exe live PID
2. il2cpp_run_dumper → automatically run Il2CppDumper from PID
3. il2cpp_scan_taskbarhero_offsets → heuristically match anchors on dump.cs:
   • Base health & Damage RVA (pj.gsi)
   • Hero Receiver Godmode RVA (pf.gsi)
   • Stat multiplier RVA (zo.haz)
   • AoE radius & hitbox RVA (bec.nax)
4. Compute Runtime Addresses (Base + RVA) and output Markdown comparison table
5. memory_write_bytes → apply godmode byte patches with confirmation

Hook a game function with GDB

"Hook the damage function at RVA 0x958ADC in GameAssembly.dll and multiply damage by 1000"

The AI will:

1. rva_disassemble_preview → inspect the function
2. gdb_hook_preview → preview the hook script
3. gdb_hook_start → attach and hook (with confirmation)
4. Monitor via log output

🛠️ Tool Reference

Process & Memory

Scanning

Writing & Allocation

GDB Hooks

IL2CPP Reverse Engineering & Automation

Workspace & Reports

Cheat Tables

🔒 Safety Design

This tool is built with defense-in-depth safety:

1. Preview First — All destructive operations have a preview/dry-run mode
2. Explicit Confirmation — Writes require confirm_write: true, hooks require confirm_hook: true
3. Write Limits — max_writes caps the number of addresses modified
4. Dry Run — dry_run: true simulates without touching memory
5. GDB Command Whitelist — Only set, printf, if/else/end allowed in hook scripts
6. Session Timeout — Scan sessions auto-expire after 30 minutes
7. Live PID Check — Validates process is alive before any operation

🏗️ Building from Source

Linux

git clone https://github.com/gede-cahya/cheat-engine-mcp.git
cd cheat-engine-mcp
cargo build --release
# Binary: ./target/release/cheat-engine-mcp

Windows

git clone https://github.com/gede-cahya/cheat-engine-mcp.git
cd cheat-engine-mcp
cargo build --release
# Binary: .\target\release\cheat-engine-mcp.exe

Install to PATH

./install.sh
# Installs to ~/.local/bin/cheat-engine-mcp

# Custom prefix:
PREFIX=/usr/local sudo -E ./install.sh

🧪 Testing

cargo fmt --check
cargo test
cargo check
cargo build --release
(cd examples/dummy-target && cargo check)

Manual MCP Test

# Initialize
printf '%s\n' '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{}}' | cargo run -q

# List tools
printf '%s\n' '{"jsonrpc":"2.0","id":2,"method":"tools/list","params":{}}' | cargo run -q

# Ping
printf '%s\n' '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"ping","arguments":{}}}' | cargo run -q

Dummy Target

# Terminal 1: Start dummy target
cd examples/dummy-target && cargo run

# Terminal 2: Scan with PID from terminal 1
printf '%s\n' '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"scanmem_scan_exact","arguments":{"pid":PID,"value":"100"}}}' | cargo run -q

🐧 Linux Troubleshooting

ptrace Permission

# Check current setting
cat /proc/sys/kernel/yama/ptrace_scope

# Allow ptrace (for development)
sudo sysctl kernel.yama.ptrace_scope=0

# Restore default (more secure)
sudo sysctl kernel.yama.ptrace_scope=1

Common Issues

📁 Project Structure

cheat-engine-mcp/
├── src/
│   ├── main.rs              # MCP server (78 tools, JSON-RPC router)
│   ├── native_scan.rs       # Pure cross-platform memory buffer scanner
│   ├── windows.rs           # Win32 API native process memory backend
│   └── il2cpp_heuristic.rs  # Heuristic anchor offset recovery engine
├── Cargo.toml               # Rust project config
├── install.sh               # Linux install script
├── install-antigravity.sh   # Antigravity one-click installer
├── install-antigravity.ps1  # Antigravity Windows installer
├── install-claude-code.sh   # Claude Code one-click installer
├── install-claude-code.ps1  # Claude Code Windows installer
├── skills/
│   └── antigravity/         # Antigravity skill definition
│       ├── SKILL.md
│       └── AGENTS.md
├── .claude/
│   └── skills/
│       └── cheat-engine-mcp/  # Claude Code skill definition
│           └── SKILL.md
├── docs/
│   ├── SETUP_ANTIGRAVITY.md
│   ├── SETUP_CLAUDE_CODE.md
│   ├── SETUP_CLAUDE_DESKTOP.md
│   ├── SETUP_CURSOR.md
│   ├── SETUP_GENERIC.md
│   ├── REVERSE_OFFSETS.md
│   ├── RULES_INSTALL.md
│   ├── RELEASE.md
│   └── *.example.json
├── examples/
│   └── dummy-target/        # Test target for scanning
├── reverse/                 # Local reverse artifacts (gitignored)
├── .cheat-tables/           # Local cheat tables (gitignored)
├── .github/workflows/       # CI + Release automation
├── CLAUDE.md                # Claude Code project rules
├── README.md
├── ROADMAP.md
└── LICENSE

📜 License

MIT License © gede-cahya

🤝 Contributing

1. Fork the repo
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Run tests (cargo fmt --check && cargo test && cargo check)
4. Commit your changes (git commit -m 'Add amazing feature')
5. Push to the branch (git push origin feature/amazing-feature)
6. Open a Pull Request