FollowAgents AI AGENTS DIRECTORY
Topic guide

REA explained: AI reverse engineering and where to start

Understand how REA connects your coding agent, MCP and analysis engines, then choose a setup guide, a project review or a related tool for your task.

Updated17 October 2026

Start here

  • REA is a reverse-engineering tool project, not a standalone language model. It gives agents inspection workflows and evidence to investigate.
  • Choose the artifact first: static JavaScript/Electron inspection and deep native analysis have different prerequisites.
  • Use this hub for orientation, our tutorial for setup, and the project profile for suitability and FARS review evidence.
  • Setup and capability guidance follows public documentation. Our REA 6.5.0 case now includes a tested static analysis of a small Electron source fixture; runtime behavior and other platforms remain untested.

What REA adds to a coding agent

REA gives an agent tools to inspect software artifacts and follow findings back to files or analysis results. Its terminal interface also supports direct investigations. The useful question is what evidence supports a conclusion, rather than how confident an explanation sounds.

This hub covers morluto/rea, published as rea-agents on npm. Check the project identity before choosing a similarly named tool.

REA, Ghidra, MCP and skills

The agent handles your request; REA coordinates inspection; an appropriate analysis engine handles the targets it supports. Ghidra is one possible native engine, not another name for REA.

MCP exposes tools to an agent. A skill supplies workflow instructions. Having the instructions installed does not establish that the live MCP connection or a native engine is ready.

Choose a workflow by artifact

  • JavaScript / Electron: start with an extracted directory or ASAR and examine entry points and module relationships.
  • Native programs: configure an engine that supports the target before inspecting functions and references.
  • Existing analysis: read its evidence and unknowns before asking for an explanation.

Requirements and host support vary by workflow and version. Use the matching documentation for commands and result contracts.

Try a real run: find an Electron CSV export channel

Download our five-file fixture and use a pinned release to inspect a preload API, IPC channel and handler. The case includes real findings and two failed attempts, with static relationships separated from runtime proof.

Connect first, then investigate one small question

Start the official setup wizard, review the selected client configuration changes and approve the intended scope. Restart the agent and check tool discovery in the new session.

For a first exercise, ask for entry points and module relationships in your own test application, with evidence and unresolved questions. This is an editorial task suggestion, not a reproduced case study.

npx rea-agents@latest setup

What “reverse engineer anything” does not establish

The project name is not a promise of universal format or platform support. Native analysis, static inspection and runtime observation have different boundaries. A successful operation does not establish complete understanding of an application.

Separate observations, inferences and unknowns. Validate important behavior with traceable evidence or your own tests. We do not infer accuracy or performance figures from the documentation.

Related tools to investigate

These are separate catalogued projects to compare by task and review evidence. Listing them here does not imply membership in the official REA ecosystem or direct interchangeability.

Where to go next

  • Choosing a tool: read the profile and source review.
  • Ready to try it: follow setup and first-use checks.
  • Checking a command or new capability: consult official documentation for your installed release.