Reverse Engineering Anything in Minutes with Claude Opus 5.5 and REA MCP
Discover the structured prompt workflow shared by @josesilesdata that combines Claude Opus 5.5 with REA MCP tools to analyze and clean-room re-implement feature
X user @josesilesdata has published a comprehensive super-prompt workflow designed to harness Claude Opus 5.5 alongside the REA (Reverse Engineer Anything) MCP server. This framework enables developers to inspect, trace, and clean-room re-implement complex features from desktop utilities, Electron apps, .NET assemblies, Android APKs, and websites within minutes.
Image source: @josesilesdata on X
Prompt Architecture and Role Definition
Rather than encouraging speculative guesswork, the super-prompt frames the AI as an elite AI reverse engineering analyst skilled in binary analysis, application architecture, runtime observation, and product engineering.
Users provide a target application and a specific feature, granting the AI authorized access through the REA MCP server (npx rea-agents setup) to leverage disassemblers like Hopper or Ghidra, static JavaScript analysis, source maps, and network inspection tools.
The Eight-Step Investigation Workflow
The prompt mandates a structured, evidence-driven process to separate proven behavior from assumptions:
- Scope Definition: Identify the target, the exact feature, and verify authorization.
- Surface Mapping: Enumerate modules, imports, symbols, strings, paths, and IPC channels using static analysis.
- Feature Tracing: Follow execution from the initial UI interaction through preload scripts, IPC bridges, and native functions.
- Reconstruction: Write clean, original code using your own tech stack based on recovered pseudocode and logic.
- Verification: Build test cases against observed inputs and outputs to confirm behavioral parity.
Evaluation Standards and Limitations
A successful re-implementation requires verifiable evidence, functional parity, transparent limitation disclosures, and clean, maintainable original code. The workflow explicitly forbids bypassing authentication, DRM, paywalls, or extracting credentials, keeping the focus strictly on interoperability, learning, and clean-room reimplementation.
Original source
- Original author: @josesilesdata on X