@enigmax/vega

Vega is an optional enigma agent built on REA (Reverse Engineer Anything): one MCP server for reverse engineering native binaries, Electron/JavaScript apps, .NET assemblies, Android APKs, firmware, websites and captures, plus REA's own reverse-engineer-anything skill and investigation commands.

This package ships only static assets. It is not a runtime dependency of enigma-cli and is not installed by default. enigma-cli fetches it on demand and runs it in an isolated agent context, so REA's 200+ tools never load into your normal coding agent.

enigma vega                  # install on first use and launch the Vega agent
enigma agent setup vega      # optional: discover Ghidra/Hopper/IDA and check REA's health

What enigma does on top of npx rea-agents setup:

• Installs the REA engine (rea-agents, pinned to the release the bundled skill matches) once and starts it with a Node.js that satisfies REA's engine range, instead of resolving npx on every agent start.
• Registers the MCP server in Claude Code, Codex, OpenCode and Kimi with long tool timeouts, so a multi-minute Ghidra analysis is not cut off by the client (morluto/rea#3509), and keeps any provider settings rea setup wrote.
• Seeds your existing login into the isolated context, so the agent starts signed in.

Inside the session: /investigate, /compare, /trace-crash, /reconstruct, /unknowns, /capture, each following one of REA's guided workflows.

Needs Node.js 22.19+, 24.11+ or 26+. Deep native analysis needs your own Ghidra, Hopper or IDA; static JavaScript and .NET analysis do not.

Use Vega only for lawful, authorized analysis - see REA's disclaimer.

The bundled skill is refreshed by the maintainer with npm run sync:vega -- --version <x.y.z> from the enigma monorepo root.

License

enigma's files: Apache-2.0. REA's skill (assets/skills/reverse-engineer-anything): MIT, copyright morluto - see LICENSE-REA.