Understanding the Three-Layer Reverse-Skill Routing Architecture

The reverse-skill routing architecture consists of three distinct layers: a platform-independent shared layer, a Windows-specific implementation layer, and a Kali Linux-specific implementation layer.

The reverse-skill project implements a modular, cross-platform routing system designed for cybersecurity workflows. At its core, this architecture separates concerns between universal logic and operating-system-specific implementations, enabling consistent skill execution across diverse environments.

The Three Layers of the Reverse-Skill Routing Architecture

The system is organized into three tiers that work together to provide platform-agnostic flexibility with OS-specific optimizations:

Layer 1: Shared (Platform-Independent) Core

The foundation of reverse-skill resides in the skills/ directory and related repositories. This layer contains:

• Skill definitions – All SKILL.md files that catalog available capabilities
• Routing matrix – skills/routing.md maps user intents to appropriate skill workflows
• CTF-Sandbox-Orchestrator sub-skills – Containerized challenge environments
• Auto-evolution systems – The field-journal/ directory records execution evidence and enables self-learning mechanisms
• Documentation generators – Tools that maintain synchronized, up-to-date skill references

This layer executes identically on any operating system. The routing engine processes skill definitions here before delegating platform-specific tasks to lower layers.

Layer 2: Windows Platform Layer

Windows-specific functionality lives in skills/scripts/ and related configuration files:

Component	Purpose
skills/scripts/*.ps1	PowerShell automation scripts
skills/bootstrap-manifest.json	Declares Winget packages and GitHub Release ZIP dependencies
skills/RULES.md	Windows-specific security constraints

The Windows layer handles self-bootstrapping of missing utilities through skills/scripts/bootstrap-reverse.ps1. When a required tool is absent, this script consults the manifest and automatically installs dependencies via Winget or direct GitHub downloads.

Layer 3: Kali Linux Platform Layer

Linux-specific implementations occupy the kali/ directory tree:

Component	Purpose
kali/scripts/*.sh	Bash automation scripts
kali/scripts/bootstrap-manifest.json	APT, pip, npm, and GitHub tar.gz package declarations
kali/RULES-kali.md	Kali-specific security hardening rules

This layer leverages Kali's native package ecosystem. The bootstrap system can install tools from multiple sources simultaneously, ensuring penetration testing environments are provisioned correctly.

How the Reverse-Skill Routing Engine Operates

The routing architecture follows a clear execution flow. First, the engine loads shared skill definitions from skills/SKILL.md and the routing matrix from skills/routing.md. Then it detects the current environment and selects the appropriate platform layer. Finally, it executes the chosen workflow with OS-specific tooling.

Platform Detection and Manifest Selection

The internal logic for choosing the correct platform layer follows this pattern:

import platform, json, pathlib

def select_manifest():
    sys = platform.system()
    if sys == "Windows":
        return pathlib.Path("skills/bootstrap-manifest.json")
    else:  # assumes Kali-compatible Linux

        return pathlib.Path("kali/scripts/bootstrap-manifest.json")

This detection occurs at runtime, ensuring the correct bootstrap manifest and script extensions load for the current operating system.

Practical Execution Examples

Cross-platform master routing (Linux/macOS/Kali):

bash skills/scripts/master-route.sh --hint "pwn-chain"

Windows-specific bootstrap execution:

powershell -NoProfile -ExecutionPolicy Bypass -File skills/scripts/bootstrap-reverse.ps1

These entry points demonstrate how the same high-level intent—executing a reverse engineering skill chain—routes through different platform layers depending on environment.

Key Architectural Files in Reverse-Skill

Understanding the routing architecture requires familiarity with these critical paths:

• skills/SKILL.md – Aggregates all skills and serves as the primary routing entry point
• skills/routing.md – Contains the intent-to-skill matching matrix used by the routing engine
• docs/ARCHITECTURE.md – Visual documentation of the three-layer structure, including the "多平台支持架构" (multi-platform support architecture) diagram
• skills/scripts/bootstrap-reverse.ps1 – Windows platform layer implementation for dependency resolution
• kali/scripts/bootstrap-manifest.json – Kali platform layer package declarations
• field-journal/ – Evidence repository enabling automatic skill evolution

As implemented in zhaoxuya520/reverse-skill, this architecture balances portability through the shared layer with optimization through platform-specific implementations. The routing engine unifies these layers into a coherent system that adapts transparently to its execution environment.

Summary

• The reverse-skill routing architecture employs three distinct layers: shared (platform-independent), Windows-specific, and Kali Linux-specific.
• The shared layer in skills/ provides universal skill definitions, routing matrices, and auto-evolution through field-journal/.
• Platform layers handle OS-specific tooling, package management (Winget for Windows, APT/pip/npm for Kali), and security rule enforcement.
• The routing engine dynamically selects the appropriate layer at runtime based on platform.system() detection.
• Bootstrap manifests (skills/bootstrap-manifest.json and kali/scripts/bootstrap-manifest.json) declaratively specify dependencies for automatic installation.

Frequently Asked Questions

How does reverse-skill detect which platform layer to use?

The system uses Python's platform.system() function to identify the operating system. If the result is "Windows", it loads skills/bootstrap-manifest.json and PowerShell scripts. For all other systems, it defaults to the Kali Linux layer, loading kali/scripts/bootstrap-manifest.json and Bash scripts.

What happens if a required tool is missing on Windows?

The skills/scripts/bootstrap-reverse.ps1 script executes automatically. It parses skills/bootstrap-manifest.json to identify missing dependencies, then installs them via Winget or direct download from GitHub Releases, requiring no manual intervention.

Can the shared layer function without platform-specific components?

Partially. The core routing logic and skill definitions in skills/SKILL.md and skills/routing.md are fully platform-agnostic. However, actual skill execution requires the appropriate platform layer for tool invocation and environment setup. The architecture is designed to degrade gracefully with clear error messaging when platform support is missing.

Where is the platform selection logic documented visually?

The "多平台支持架构" (multi-platform support architecture) diagram in docs/ARCHITECTURE.md illustrates how the three layers interact. This documentation shows the routing engine's position between shared skill definitions and platform-specific implementations.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}