The REA MCP reverse engineering (Reverse Engineer Anything) open-source project is gaining attention among developers and security researchers. It acts as a Model Context Protocol (MCP) server that connects coding agents — such as Claude Code, Cursor, Codex and Gemini CLI — to classic reverse-engineering tools including Ghidra, Hopper and IDA Pro.
The core idea is straightforward: instead of a human manually switching between disassemblers, decompilers and terminals, the AI agent can request structured analyses, receive results with evidence, confidence and explicit limitations, and continue the investigation. Everything runs locally. REA does not replace the analysis engines; it orchestrates and standardizes access to them.
What REA can analyze
The project, hosted at github.com/morluto/rea and documented at rea.tools, covers a wide range of targets:
- Native binaries (Mach-O, ELF, PE) via Hopper, Ghidra or IDA Pro — returning pseudocode, assembly, strings, symbols and cross-references.
- JavaScript and Electron applications (including ASAR).
- .NET assemblies.
- Android APKs (with headless JADX support).
- Websites and network captures.
- Firmware (via Binwalk/Unblob in some cases).
Basic installation is done with a single command: npx rea-agents setup. This registers the MCP server with the chosen agent and installs workflow skills. Native analysis requires having (or installing) one of the supported providers. Static JavaScript analysis needs no additional engine.
Why it matters
Traditional reverse engineering demands expertise and time to navigate complex tools. With AI agents able to orchestrate these tools and present findings with traceable evidence, the barrier lowers for analysis tasks, understanding features in apps without source code, security research and even reconstructing behaviors. The project emphasizes that results include limitations and do not promise full recovery of original source code, especially in obfuscated binaries.
The repository surpassed tens of thousands of stars in October 2026 and continues to receive active commits. This reflects growing interest in tools that integrate language models into specialized technical workflows rather than mere prompt-based code generation.
Sources
Transparency: This content was created, edited or reviewed with the assistance of artificial intelligence. The information was cross-checked with public posts on X and available internet sources. Consult the original sources for full context.
Por GeekikiBot