What is REA MCP?
MCP is the interface through which an agent calls tools. REA exposes analysis capabilities to agents that support local MCP servers. The agent asks questions and interprets results. Adding a registration does not install every reverse-engineering engine or prove a target has been analyzed.
Agent client → REA MCP server → target-specific tools → checkable results
Check three layers separately
- Client: configure the agent you actually use and restart it afterward.
- MCP: the agent discovers REA tools and doctor reports no unresolved problem for the selected registration.
- Target: required runtimes, engines, paths and input are available; inspect the output of a minimal task.
Choose dependencies by target
- Static JavaScript: a readable directory or supported ASAR; Ghidra is not required.
- Native binaries: prepare a compatible Ghidra, Hopper or IDA installation and check the engine itself.
- Browser behavior: prepare a dedicated Chromium debugging session; a webpage URL is not its debugging endpoint.
Which client should you configure?
Use the setup assistant for Claude Code, Codex or Cursor. Choose your existing agent and environment, inspect the dry run, then apply the selected setup. You do not need to configure every detected client to test one integration.
Does “reverse engineering MCP” mean only REA?
No. It is also a generic search for MCP integrations around reverse-engineering tools. Compare target support, dependencies, result formats and evidence, rather than just tool names or counts. This page explains the REA approach.
Continue your investigation
- Build a setup plan for your environment
- Inspect the reproducible case study
- Troubleshoot by symptom
- Browse all field guides
Based on upstream REA documentation, reviewed October 17, 2026. Documentation-based guidance is not a live test of every client or target. Official setup reference · Project source and overview