
cheatengine-mcp-automation
Automate Cheat Engine memory analysis, reverse engineering, and debugging using AI through MCP protocol
- Installs
- 207
- Last 24h
- —
- Trending rank
- —
- Security
- Not audited
Install
Works with every agent the skills CLI supports.
npx skills add https://github.com/reason-machines/mcp-skills --skill cheatengine-mcp-automation- 01
Install
Run the command in your project. The CLI asks which agents to install for.
- 02
Describe the task
Ask your agent for something this skill covers. It loads cheatengine-mcp-automation automatically when relevant.
- 03
Review the result
The agent follows the SKILL.md steps, scripts and references bundled with the skill.
SKILL.md
The instructions your agent reads when the skill is triggered.
Cheat Engine MCP Automation
Skill by ara.so — MCP Skills collection.
This skill enables AI agents to control Cheat Engine through the Model Context Protocol, automating memory analysis, pointer scanning, structure dissection, and reverse engineering tasks that normally take days or weeks.
What It Does
Connect Claude, Cursor, or any MCP-compatible AI to Cheat Engine for:
- Memory scanning: Find values (int, float, string, AOB patterns) across gigabytes instantly
- Pointer chain resolution: Follow
[[base+0x10]+0x20]+0x8paths automatically - Structure analysis: Auto-detect field types and offsets in memory structures
- Code disassembly: Analyze functions, find references, identify C++ RTTI classes
- Invisible debugging: Hardware breakpoints + Ring -1 DBVM tracing
- Code injection: Allocate memory, inject DLLs, execute shellcode
- Process automation: Attach, pause, resume, create processes programmatically
Installation
Prerequisites
- Windows only (uses Named Pipes)
- Cheat Engine 7.5+ installed
- Python 3.10+
1. Install Python Dependencies
cd MCP_Server
pip install -r requirements.txt
Or manually:
pip install mcp pywin32
2. Configure MCP Server
Add to your MCP client config (location varies by client):
Claude Desktop (~/AppData/Roaming/Claude/claude_desktop_config.json):
{
"mcpServers": {
"cheatengine": {
"command": "python",
"args": ["C:/path/to/cheatengine-mcp-bridge/MCP_Server/mcp_cheatengine.py"]
}
}
}
Cursor (.cursor/mcp.json in project):
{
"servers": {
"cheatengine": {
"command": "python",
"args": ["C:/absolute/path/to/MCP_Server/mcp_cheatengine.py"]
}
}
}
Codex (~/.codex/config.toml):
[mcp_servers.cheatengine]
command = "python"
args = ['C:\path\to\cheatengine-mcp-bridge\MCP_Server\mcp_cheatengine.py']
3. Load Bridge in Cheat Engine
Open Cheat Engine, then load the Lua bridge:
Option A: Execute Script (Recommended)
- Go to
File→Execute Script - Select
MCP_Server/ce_mcp_bridge.lua - Click
Execute
Option B: Lua Engine
- Open
Table→Show Cheat Table Lua Script - Paste this line:
dofile([[C:\path\to\cheatengine-mcp-bridge\MCP_Server\ce_mcp_bridge.lua]])
- Execute
Look for confirmation:
[MCP v12.0.0] MCP Server Listening on: CE_MCP_Bridge_v99
4. Verify Connection
Restart your MCP client, then use the ping tool:
User: "Ping cheat engine"
Expected response:
{
"success": true,
"version": "12.0.0",
"message": "CE MCP Bridge Active",
"process_id": 0
}
Critical Configuration
Prevent BSODs
MUST DO: Disable this setting to prevent CLOCK_WATCHDOG_TIMEOUT blue screens:
- Cheat Engine →
Edit→Settings - Go to
Extratab - UNCHECK "Query memory region routines"
- Restart Cheat Engine
This prevents conflicts between memory scanning and DBVM/anti-cheat systems.
Enable DBVM (Optional)
For Ring -1 debugging and invisible breakpoints:
- Cheat Engine →
Kernel Tools→DBVM - Follow prompts to install hypervisor
- Reboot system
- DBVM functions (
start_dbvm_watch, etc.) now available
Core MCP Tools
Process Management
-- List running processes
get_process_list()
-- Returns: [{pid, name, windowTitle}, ...]
-- Attach to process
open_process({process_name = "game.exe"})
-- Or by PID: open_process({process_id = 1234})
-- Get current process info
get_process_info()
-- Returns: {name, pid, base_address, entry_point}
-- Launch new process under CE control
create_process({path = "C:/game/game.exe", parameters = "--debug"})
-- Pause/resume execution
pause_process()
unpause_process()
Memory Reading
-- Read typed values
read_integer({address = "0x400000", size = 4}) -- 4-byte int
read_float({address = "0x400000", double = true}) -- 8-byte double
read_string({address = "0x400000", length = 100}) -- Read 100 chars
read_bytes({address = "0x400000", size = 16}) -- Raw bytes
-- Follow pointer chains
read_pointer_chain({
base = "game.exe",
offsets = [0x123456, 0x10, 0x20, 0x8]
})
-- Resolves: [[game.exe+0x123456]+0x10]+0x20]+0x8
-- Read structures
read_memory({address = "0x400000", size = 256})
Memory Scanning
-- Scan for specific value
scan_all({
value = "15000",
value_type = 4, -- 4-byte int
writable = true
})
-- Returns: [{address, value}, ...]
-- Next scan (filter previous results)
scan_all({
value = "15100",
value_type = 4,
next_scan = true
})
-- AOB (Array of Bytes) pattern scanning
aob_scan({
pattern = "48 8B 05 ?? ?? ?? ?? 48 85 C0",
writable = false,
executable = true
})
-- Find what writes to address
find_what_writes({address = "0x400000"})
-- Returns: [{instruction, address}, ...]
Code Analysis
-- Disassemble function
disassemble({
address = "0x401000",
count = 20 -- Instructions to disassemble
})
-- Analyze function structure
analyze_function({address = "0x401000"})
-- Returns: {prologue, epilogue, calls, jumps, stack_size}
-- Find all references to address
find_references({address = "0x500000"})
-- Find all CALL instructions to function
find_call_references({address = "0x401000"})
-- Get C++ class name via RTTI
get_rtti_classname({address = "0x600000"})
-- Returns: "CPlayerInventory"
Structure Dissection
-- Auto-analyze memory structure
dissect_structure({
address = "0x500000",
size = 512,
name = "PlayerData"
})
-- Returns: [
-- {offset: 0x00, type: "ptr", name: "vtable", value: "0x1234"},
-- {offset: 0x08, type: "int", name: "health", value: "100"},
-- {offset: 0x0C, type: "float", name: "x_pos", value: "123.45"}
-- ]
Debugging & Breakpoints
-- Hardware breakpoint (execution)
set_breakpoint({
address = "0x401000",
type = 1 -- 1=execute, 2=write, 3=access
})
-- Data breakpoint (watch memory)
set_data_breakpoint({
address = "0x500000",
size = 4,
on_write = true
})
-- DBVM invisible tracing (Ring -1)
start_dbvm_watch({
address = "0x401000",
watch_writes = true
})
-- Remove breakpoint
remove_breakpoint({address = "0x401000"})
Code Injection
-- Allocate memory in target
allocate_memory({size = 4096, near = "game.exe"})
-- Returns: {address: "0x10000000"}
-- Inject DLL
inject_dll({dll_path = "C:/mods/mymod.dll"})
-- Assemble instruction
assemble_instruction({
instruction = "mov rax, [rbx+0x10]",
address = "0x400000" -- For relative addressing
})
-- Returns: {bytes: "48 8B 43 10"}
-- Execute shellcode
execute_code({
code = "90 90 C3", -- nop nop ret
address = "0x10000000"
})
-- Generate API hook template
generate_api_hook_script({
function_address = "0x401000",
function_name = "ProcessPacket"
})
Symbol Management
-- Register named symbol
register_symbol({
name = "PlayerBase",
address = "0x500000"
})
-- Use in other commands
read_pointer_chain({
base = "PlayerBase",
offsets = [0x10, 0x20]
})
-- Get symbol info
get_symbol_info({name = "PlayerBase"})
-- Enable Windows PDB symbols
enable_windows_symbols()
Common Patterns
Finding Player Health
User: "Find player health, currently at 100"
AI workflow:
1. scan_all({value: "100", value_type: 4})
2. User changes health to 95
3. scan_all({value: "95", value_type: 4, next_scan: true})
4. find_what_writes({address: first_result})
5. disassemble({address: write_instruction})
6. analyze_function({address: function_start})
Tracing Packet Encryption
User: "Find where network packets are encrypted"
AI workflow:
1. aob_scan({pattern: "E8 ?? ?? ?? ?? 48 8B", executable: true})
// Common CALL pattern before crypto
2. For each result:
- disassemble({address: result, count: 30})
- find_call_references({address: called_function})
3. set_breakpoint({address: suspect_function, type: 1})
4. start_dbvm_watch({address: suspect_function})
Understanding C++ Object
User: "What is the object at [[game.exe+0x123456]+0x10]?"
AI workflow:
1. read_pointer_chain({base: "game.exe", offsets: [0x123456, 0x10]})
2. get_rtti_classname({address: result}) // "CPlayerInventory"
3. dissect_structure({address: result, size: 512, name: "CPlayerInventory"})
4. For each pointer field:
- read_pointer_chain({base: result, offsets: [field_offset]})
- get_rtti_classname({address: pointer_value})
Creating Update-Proof AOB
User: "Find a unique pattern for the health function"
AI workflow:
1. find_what_writes({address: health_address})
2. disassemble({address: write_instruction, count: 50})
3. Analyze for unique opcodes (avoid relative offsets)
4. aob_scan({pattern: "48 8B 05 ?? ?? ?? ?? 48 85 C0 74 ?? 8B 40 08"})
5. Verify single result across full memory space
Automating Pointer Path Discovery
-- Example: Find pointer path from base to target
User: "Find the pointer chain from game.exe to 0x12345678"
AI workflow:
1. get_process_info() // Get base address
2. scan_all({value: "0x12345678", value_type: 8}) // Find pointers to target
3. For each pointer result:
- scan_all({value: pointer_result, value_type: 8})
4. Build chain: base -> ptr1 -> ptr2 -> target
5. register_symbol({name: "TargetPath", address: "[[game.exe+0xABC]+0x10]+0x20"})
Real Example: Complete Game Trainer Workflow
User: "Create a god mode for MyGame.exe"
AI session:
───────────────────────────────────────────────────
You: Attach to MyGame.exe
AI: open_process({process_name: "MyGame.exe"})
✓ Attached to PID 5432, base: 0x140000000
You: Find my health, currently 100
AI: scan_all({value: "100", value_type: 4})
→ Found 2,847 results
You: Took damage, now at 87
AI: scan_all({value: "87", value_type: 4, next_scan: true})
→ Filtered to 3 addresses
You: What writes to the first one?
AI: find_what_writes({address: "0x20A3B4C0"})
→ MyGame.exe+0x4F2A10 - mov [rax+08],edx
You: Disassemble that function
AI: disassemble({address: "0x1404F2A00", count: 40})
→ Found: Function processes damage, compares against max health at rax+0C
You: What's the full structure?
AI: dissect_structure({address: "0x20A3B4C0", size: 256})
→ +00: vtable
→ +08: current_health (int) = 87
→ +0C: max_health (int) = 100
→ +10: shield (int) = 0
→ +14: invulnerable (byte) = 0
You: Create a cheat to set invulnerable to 1
AI: 1. register_symbol({name: "PlayerHealth", address: "0x20A3B4C0"})
2. Generated script:
[ENABLE]
PlayerHealth+14:
db 01
[DISABLE]
PlayerHealth+14:
db 00
You: Make it work after game updates
AI: aob_scan({pattern: "48 8B 05 ?? ?? ?? ?? 48 85 C0 74 ?? 89 50 08"})
→ Found unique pattern at MyGame.exe+0x4F2A0E
Updated script with AOB injection point
Environment Variables
Set these before starting the MCP server:
# Increase timeout for complex operations (default: 30s)
set CE_MCP_TIMEOUT=60
# Enable shell commands (SECURITY RISK - only for trusted automation)
set CE_MCP_ALLOW_SHELL=1
Troubleshooting
"Too many local variables" error in Cheat Engine
Solution: Use dofile() instead of pasting the script:
dofile([[C:\path\to\ce_mcp_bridge.lua]])
The bridge uses global functions to avoid Lua's 200 local variable limit.
MCP client can't connect
Check:
- Cheat Engine shows
MCP Server Listening on: CE_MCP_Bridge_v99 - MCP client restarted after config change
- Python path in config is absolute (not relative)
pip install mcp pywin32completed successfully- Run
pingtool - should returnsuccess: true
process_id: 0 in ping response
Normal - means no process attached yet. Use open_process() first.
BSOD (CLOCK_WATCHDOG_TIMEOUT)
Cause: "Query memory region routines" enabled + DBVM conflict
Fix:
- Cheat Engine → Settings → Extra
- UNCHECK "Query memory region routines"
- Restart Cheat Engine
Commands return "CE not attached"
Solution: Attach to a process first:
open_process({process_name: "target.exe"})
DBVM functions fail
Check:
- DBVM installed: Cheat Engine → Kernel Tools → DBVM
- System rebooted after DBVM install
- No conflicting hypervisors (Hyper-V, VMware)
Testing Your Setup
Run the test suite to verify all ~180 tools work:
cd MCP_Server
python test_mcp.py
Expected output:
✅ Memory Reading: 6/6 tests passed
✅ Process Info: 4/4 tests passed
✅ Code Analysis: 8/8 tests passed
✅ Breakpoints: 4/4 tests passed
✅ DBVM Functions: 3/3 tests passed
✅ Utility Commands: 11/11 tests passed
────────────────────────────────────
Total: 36/37 PASSED (100% success)
Key Value Types Reference
-- value_type parameter for scanning:
1 = byte (1 byte)
2 = 2 bytes
4 = 4 bytes (int32)
8 = 8 bytes (int64)
5 = float (4 bytes)
6 = double (8 bytes)
7 = string
8 = pointer (4 or 8 bytes depending on process)
9 = AOB (array of bytes)
Architecture Overview
AI Agent (Claude/Cursor/Codex)
↓ JSON-RPC over stdio
Python MCP Server (mcp_cheatengine.py)
↓ Named Pipe: \\.\pipe\CE_MCP_Bridge_v99
Lua Bridge (ce_mcp_bridge.lua)
↓ Cheat Engine API
Target Process Memory
The Python server translates MCP protocol to pipe commands, the Lua bridge executes them using Cheat Engine's API, and results flow back through the same chain.
Security Note
This tool enables arbitrary code execution in target processes. Only use on:
- Single-player games you own
- Your own software for testing
- CTF/research environments
- Educational reverse engineering
Never use on:
- Multiplayer games (violates ToS, ruins others' experience)
- Software you don't have authorization to modify
- Production systems
FAQ
What does the cheatengine-mcp-automation skill do?
Automate Cheat Engine memory analysis, reverse engineering, and debugging using AI through MCP protocol
How do I install cheatengine-mcp-automation?
Run npx skills add https://github.com/reason-machines/mcp-skills --skill cheatengine-mcp-automation in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/cheatengine-mcp-automation/ (personal) or .claude/skills/cheatengine-mcp-automation/ (project).
Which AI agents can use cheatengine-mcp-automation?
Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.
Is cheatengine-mcp-automation safe to install?
No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.
Who maintains cheatengine-mcp-automation?
It is published by reason-machines in reason-machines/mcp-skills. It has 207 recorded installs.
More from reason-machines/mcp-skills
codebase-memory-mcp-intelligence
reason-machines/mcp-skills
High-performance code intelligence MCP server that indexes codebases into knowledge graphs for structural queries, call traces, and architecture analysis
playwright-mcp-server
reason-machines/mcp-skills
Open the skill to read its SKILL.md instructions.
tradingview-mcp-assistant
reason-machines/mcp-skills
Open the skill to read its SKILL.md instructions.
lanhu-mcp-collaboration
reason-machines/mcp-skills
Open the skill to read its SKILL.md instructions.
github-mcp-server
reason-machines/mcp-skills
Open the skill to read its SKILL.md instructions.
google-meta-ads-ga4-mcp
reason-machines/mcp-skills
Open the skill to read its SKILL.md instructions.
Popular ai & agents skills
google-agents-cli-publish
google/agents-cli
This skill should be used when the user wants to "publish an agent", "publish my ADK agent", "register an agent with Gemini", "publish to Gemini" (formerly Gemini Enterprise), or needs guidance on the agents-cli publish gemini command. Also use when the user wants to "manage agents in Agent Registry", "list/update/delete registered agents", or "register an MCP server". Covers ADK vs A2A registration modes, programmatic and interactive usage, flag reference, auto-detection from deployment metadata, Agent Registry fleet management, and troubleshooting. Part of the agents-cli skills suite. Do NOT use for deployment (use google-agents-cli-deploy).
google-agents-cli-eval
google/agents-cli
This skill should be used when the user wants to "run an evaluation", "evaluate my agent", "evaluate my ADK agent", "write an eval dataset", "analyze eval failures", "compare eval results", "optimize agent", or needs guidance on the Gemini platform eval methodology and the Quality Flywheel. Covers eval metrics, dataset schema, LLM-as-judge scoring, and common failure causes. Applies to any agents-cli project, whatever framework the agent is written in. Do NOT use for agent API code patterns (ADK: use google-agents-cli-adk-code), deployment (use google-agents-cli-deploy), or project scaffolding (use google-agents-cli-scaffold).
skill-creator
anthropics/skills
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
codex-pet
prime-skills/runcomfy-agent-skills
Codex Pet generator on RunComfy. Build a Codex-compatible Codex Pet spritesheet.webp + pet.json from a single reference image, drop it into ${CODEXHOME:-$HOME/.codex}/pets/ / and Codex picks it up as a custom Codex Pet next to the 8 built-ins. This skill produces the exact Codex Pet atlas Codex expects (1536x1872 PNG/WebP, 8 cols x 9 rows, 192x208 cells, 9 animation states — idle, running-right, running-left, waving, jumping, failed, waiting, running, review). Calls OpenAI GPT Image 2 edit ONCE via the local RunComfy CLI as runcomfy run openai/gpt-image-2/edit to produce a canonical Codex Pet pose, then assembles all 9 animation rows programmatically with ImageMagick micro-transforms — no Codex Pro, no $imagegen, no OPENAIAPIKEY required, only RUNCOMFYTOKEN. Triggers on "codex pet", "create codex pet", "make codex pet", "hatch codex pet", "/hatch image", "desktop pet codex", "codex pets", "spritesheet.webp", or any explicit ask to build a custom pet for OpenAI Codex.
full-output-enforcement
leonxlnx/taste-skill
Overrides default LLM truncation behavior. Enforces complete code generation, bans placeholder patterns, and handles token-limit splits cleanly. Apply to any task requiring exhaustive, unabridged output.
executing-plans
obra/superpowers
Use when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available