lingbol088-spec/reverse-flow-skill

What it solves

This project provides a structured reverse engineering workflow for AI Agents (specifically for Codex), enabling them to handle CTF (Capture The Flag) challenges, crackmes, and wargames in local sandbox environments. It transforms the often chaotic process of binary analysis into a stable, step-by-step pipeline.

How it works

The skill uses a combination of English internal prompts for stability and Chinese for user interaction. It guides the AI agent through a predefined sequence: Analysis $\rightarrow$ Report $\rightarrow$ Reverse Engineering $\rightarrow$ Deep Reverse Engineering $\rightarrow$ Vulnerability Assessment $\rightarrow$ User Decision. It also includes a "normalization" layer that translates colloquial user requests (like "get the flag" or "bypass anti-debug") into specific technical goals such as locating validation branches or analyzing check routines.

Who it’s for

It is designed for security researchers, CTF players, and students working in authorized local sandboxes or training ranges who want to leverage AI agents to automate and systematize the reverse engineering process.

Highlights

• Structured Pipeline: Enforces a rigorous flow from initial triage to final vulnerability judgment.
• Intent Normalization: Maps vague user requests to concrete reverse engineering tasks.
• Hybrid Language Approach: Uses English for core logic and Chinese for the user interface to maximize model reliability.
• Local-First Focus: Optimized for local CTF and crackme environments, removing the need for repetitive context setting.