IL2CPP Frida MCP Server
Enables AI assistants to analyze and manipulate Unity IL2CPP applications via Frida, supporting disassembly, class/method listing, and heap analysis.
README
IL2CPP Frida MCP Server
A Model Context Protocol (MCP) based IL2CPP reverse engineering tool that enables AI assistants (like Claude, Kiro) to directly analyze and manipulate Unity IL2CPP applications.
Features
- 🔌 Frida Integration - Support for USB, remote, and local device connections
- 📦 IL2CPP Analysis - List images, classes, methods with fuzzy search support
- 🔍 Disassembly - Method disassembly based on Frida Instruction API
- 🧠 GC Heap Analysis - Find runtime object instances
- 📤 Import/Export - Find module import and export functions
- 🛠️ JS Execution - Execute arbitrary JavaScript code to manipulate Frida API
Project Structure
.
├── mcp_server.py # Entry point
├── mcp_server/ # MCP Server module
│ ├── __init__.py
│ ├── server.py # MCP Server core
│ ├── state.py # Frida state management
│ ├── tools.py # MCP tool definitions
│ ├── transport.py # Transport layer (stdio/sse/http)
│ ├── cli.py # Command line interface
│ ├── agent_loader.py # Agent loader
│ └── handlers/ # Tool handlers
│ ├── __init__.py
│ ├── frida_handlers.py
│ └── il2cpp_handlers.py
├── agent/ # Frida Agent (TypeScript)
│ ├── index.ts # Agent entry point
│ ├── core/ # Core modules
│ │ ├── il2cpp-helper.ts
│ │ ├── method-utils.ts
│ │ └── class-finder.ts
│ └── services/ # Service modules
│ ├── image-service.ts
│ ├── class-service.ts
│ ├── method-service.ts
│ ├── disasm-service.ts
│ ├── module-service.ts
│ ├── gc-service.ts
│ └── exec-service.ts
├── _agent.js # Compiled Agent
├── package.json
├── tsconfig.json
└── requirements.txt
Installation
Option 1: pip install (Recommended)
# Install from source
pip install .
# Or install in development mode
pip install -e .
Option 2: Manual dependency installation
pip install -r requirements.txt
Compile Frida Agent
npm install
npm run build
Usage
Start MCP Server
# If installed via pip
il2cpp-frida-mcp # Interactive selection
il2cpp-frida-mcp --stdio # stdio mode
il2cpp-frida-mcp --sse # SSE mode
il2cpp-frida-mcp --http # HTTP mode
# Or run script directly
python mcp_server.py --stdio
# Custom host and port
il2cpp-frida-mcp --sse --host 0.0.0.0 --port 9000
Configure MCP Client
Claude Desktop / Kiro
Add to your MCP configuration file:
{
"mcpServers": {
"il2cpp-frida": {
"command": "il2cpp-frida-mcp",
"args": ["--stdio"]
}
}
}
Or run using Python module:
{
"mcpServers": {
"il2cpp-frida": {
"command": "python",
"args": ["-m", "mcp_server", "--stdio"]
}
}
}
MCP Tools
Frida Basic Tools
| Tool | Description |
|---|---|
frida_list_devices |
List all available Frida devices |
frida_connect |
Connect to device and target process |
frida_disconnect |
Disconnect Frida connection |
frida_resume |
Resume suspended process |
frida_list_processes |
List processes on device |
IL2CPP Analysis Tools
| Tool | Description |
|---|---|
il2cpp_list_images |
List all IL2CPP images |
il2cpp_list_classes |
List all classes in specified image |
il2cpp_list_methods |
List all methods in specified class |
il2cpp_show_method |
Show method details |
il2cpp_find_classes |
Find classes (fuzzy match supported) |
il2cpp_find_methods |
Find methods (fuzzy match supported) |
il2cpp_show_asm |
Disassemble method |
il2cpp_find_export |
Find export functions |
il2cpp_find_import |
Find import functions |
il2cpp_exec_js |
Execute arbitrary JavaScript code |
il2cpp_gc_choose |
Find instances of specified class in heap |
il2cpp_gc_info |
Get GC heap information |
Examples
1. Connect to Device
Use frida_connect to connect to the frontmost app on USB device
2. Analyze IL2CPP
List all images, then find classes containing "Player"
3. View Method Details
List all methods of PlayerController class, then view Update method details
Development
Compile Agent
npm run build # Single build
npm run watch # Watch mode
Dependencies
- Python 3.10+
- Node.js 16+
- Frida 17+
- frida-il2cpp-bridge
Acknowledgements
- frida-il2cpp-bridge - IL2CPP runtime bridge
- Il2CppHookScripts - Reference implementation
License
MIT
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。