Original post
What it actually is
bash
npx rea-agents setup
Pick your agents, approve the config changes (it backs up existing configs), restart the agent. Requires Node.js 22.x/24.x/26.x + npm. On Linux you also need a procps-compatible ps on PATH. Native analysis optionally uses an existing Hopper, Ghidra, IDA, or Kuna install.
Power move: just ask the agent
"Understand how search works in [app], show me the evidence, and build a similar feature for my project."
Terminal quick hits
bash
# Inspect an extracted JS/Electron app or ASAR
npx -y rea-agents@latest analyze-javascript-application /abs/path/to/app --json
# Install the CLI for regular use
npm install --global rea-agents
rea --help
# Keep it current (fast-moving project)
rea update
What it covers (the cheat sheet)
Category Targets
Native binaries Pseudocode, assembly, strings, symbols, calls (Hopper/Ghidra/IDA/Kuna)
JS / Electron Modules, imports, source maps, IPC, native add-ons
.NET Metadata, CIL, native deps, build comparisons (static only)
Mobile Android APKs (JADX), signatures, resources, Flutter, IL2CPP, Godot PCK
Firmware / formats Firmware regions (Binwalk/Unblob), Go, Rust, WASM, EVM bytecode, ELF/PE/Mach-O
Runtime Process behavior, network captures (HAR/SAZ/pcap), packets, logic signals, crash dumps
AI artifacts GGUF/safetensors model metadata, CUDA/HIP kernels
Why it's powerful
Evidence-first — every conclusion ships with the proof, so you're not trusting black-box guesses.
Agent-native — turns "I wonder how X works" into a traceable investigation the agent drives.
End-to-end — from observe behavior → find the code → explain it → reimplement & test your own version.
RL environments — record a web app session and export a runnable env an AI agent can practice in.
Hidden link: log in and reply to this thread to see it- One MCP for reverse engineering across binaries, applications, and runtime behavior.
- You point your agent (Claude Code, Codex, Cursor, Gemini CLI, Grok, etc.) at a target and ask a question — REA gives the agent tools to inspect it and returns findings with the evidence and limitations behind each conclusion.
- Analysis runs locally. Static JS/.NET inspection needs no native engine.
- Works from the terminal too, not just via an agent.
bash
npx rea-agents setup
Pick your agents, approve the config changes (it backs up existing configs), restart the agent. Requires Node.js 22.x/24.x/26.x + npm. On Linux you also need a procps-compatible ps on PATH. Native analysis optionally uses an existing Hopper, Ghidra, IDA, or Kuna install.
Power move: just ask the agent
"Understand how search works in [app], show me the evidence, and build a similar feature for my project."
Terminal quick hits
bash
# Inspect an extracted JS/Electron app or ASAR
npx -y rea-agents@latest analyze-javascript-application /abs/path/to/app --json
# Install the CLI for regular use
npm install --global rea-agents
rea --help
# Keep it current (fast-moving project)
rea update
What it covers (the cheat sheet)
Category Targets
Native binaries Pseudocode, assembly, strings, symbols, calls (Hopper/Ghidra/IDA/Kuna)
JS / Electron Modules, imports, source maps, IPC, native add-ons
.NET Metadata, CIL, native deps, build comparisons (static only)
Mobile Android APKs (JADX), signatures, resources, Flutter, IL2CPP, Godot PCK
Firmware / formats Firmware regions (Binwalk/Unblob), Go, Rust, WASM, EVM bytecode, ELF/PE/Mach-O
Runtime Process behavior, network captures (HAR/SAZ/pcap), packets, logic signals, crash dumps
AI artifacts GGUF/safetensors model metadata, CUDA/HIP kernels
Why it's powerful
Evidence-first — every conclusion ships with the proof, so you're not trusting black-box guesses.
Agent-native — turns "I wonder how X works" into a traceable investigation the agent drives.
End-to-end — from observe behavior → find the code → explain it → reimplement & test your own version.
RL environments — record a web app session and export a runnable env an AI agent can practice in.
