REA Releases MCP Integrating Reverse Engineering Tools

The GitHub project REA (Reverse Engineer Anything) has integrated various analysis tools into a single MCP server and CLI, enabling coding agents to investigate executables, JS/Electron apps, .NET applications, websites, and more without source code.

Key Features and Integrations

• Multi-Engine Support: Native binaries are analyzed using a selected engine among Hopper, Ghidra, or IDA, while JS/.NET applications can undergo static analysis without separate engines.
• Agent Integration: Register with 17 clients, including Claude Code, Codex, and Cursor, using the npx rea-agents setup command.
• Structured Results: Assigns an Evidence ID to all analysis results, clearly distinguishing observed facts from inferences.

Supported Targets and Tech Stack

• Supported Formats: Native binaries, JavaScript/Electron, websites, .NET assemblies, Android APKs, firmware, processes, ELF, EVM bytecode, etc.
• Environment Requirements: A TypeScript-based npm package rea-agents requiring Node.js 22.19+, 24.11+, or 26+.
• Performance: Actual app analysis results can reach hundreds of MBs (e.g., Obsidian at 334MB), but a summary view of approximately 10KB is provided.

Security and Licensing

• Analysis is performed locally, but results are transmitted to the agent model provider; standalone CLI usage is recommended in environments with restricted data transmission.
• REA itself follows the MIT License, while connected tools like Hopper and IDA are subject to their respective commercial licenses.