How reverse-skill's Multi-Platform Script Architecture Works: PowerShell vs Bash

reverse-skill uses a layered architecture that separates platform-agnostic skill definitions from thin platform-specific wrapper scripts, enabling the same reverse-engineering logic to run on Windows via PowerShell and Kali Linux via Bash.

The reverse-skill project by zhaoxuya520 is designed to run security and reverse-engineering workflows across heterogeneous environments. Rather than maintaining two separate codebases, the repository implements a single-source-of-truth pattern where skill definitions remain identical while execution mechanics adapt to the host operating system. This article examines how the multi-platform script architecture balances portability with native integration.

Platform-Specific Layer Structure

The architecture divides responsibilities across two parallel directory structures. According to the source code in docs/ARCHITECTURE.md, the platform detection and routing logic lives at the top level, while implementation details descend into environment-specific folders.

Windows PowerShell Implementation

On Windows, reverse-skill leverages PowerShell modules stored in skills/scripts/*.ps1. The bootstrap and tool resolution pipeline centers on:

• bootstrap-manifest.json – Defines installation sources using winget or direct GitHub ZIP downloads
• skills/scripts/lib/ToolDiscovery.ps1 – Resolves tool availability, checks tool-index.md, and invokes bootstrap-reverse.ps1 when dependencies are missing
• RULES.md – Platform-specific rule set that drives AI routing decisions

Kali Linux Bash Implementation

For Kali Linux environments, native Bash scripts in kali/scripts/*.sh provide equivalent functionality:

• kali/scripts/bootstrap-manifest.json – Uses apt, pip, npm, and other Linux package managers
• kali/scripts/bootstrap-reverse.sh – Direct bootstrap invocation (called via functions like ensure_tool in recon.sh)
• kali/RULES-kali.md – Kali-specific rule set with identical semantic structure to Windows rules

The shared skills/ directory contains platform-agnostic skill definitions including SKILL.md, routing.md, and reference documentation accessed by both implementations.

Execution Flow: From Platform Detection to Skill Execution

The multi-platform support architecture follows a six-stage pipeline documented in docs/ARCHITECTURE.md (lines 70-91):

1. Detect platform – The system selects RULES.md for Windows or kali/RULES-kali.md for Kali (lines 6-10)

2. Load scripts – PowerShell (*.ps1) or Bash (*.sh) implementations are selected based on the detected environment

3. Check tool index – tool-index.md is consulted to determine which tools are already installed

4. Bootstrap if needed – Missing tools trigger platform-specific installation:

   • PowerShell: ToolDiscovery.ps1 → bootstrap-reverse.ps1
   • Bash: Direct call to bootstrap-reverse.sh (e.g., ensure_tool function in recon.sh, lines 41-50)

5. Execute the skill – The platform-specific script runs identical logical steps (e.g., invoking rabin2 with optional r2 analysis)

6. Generate output – Results feed into shared documentation generators (docs-generator, diagram-generator)

Code Comparison: Radare2 Reconnaissance Skill

The Radare2 recon skill demonstrates architectural parity across platforms. Both implementations perform the same operations with language-specific syntax.

Windows PowerShell

# skills/radare2/scripts/recon.ps1

.\skills\radare2\scripts\recon.ps1 -TargetPath C:\path\to\binary.exe `
    -StringsLimit 50 -ImportsLimit 100 -RunAnalysis

Kali Linux Bash

# skills/radare2/scripts/recon.sh

bash skills/radare2/scripts/recon.sh /path/to/binary \
    --strings-limit 50 --imports-limit 100 --analyze

Both scripts execute identical analysis steps:

• Verify rabin2 availability (and optionally r2), bootstrapping if absent
• Print basic file information, sections, imports, exports, and strings
• Perform function/entry analysis when the analyze flag is set

The divergence is strictly superficial: argument parsing differs (PowerShell named parameters vs. Bash long options) and bootstrap invocation paths vary by platform convention.

Key Architectural Files

File	Platform	Responsibility
skills/scripts/lib/ToolDiscovery.ps1	Windows	Tool resolution and auto-bootstrap orchestration
kali/scripts/bootstrap-reverse.sh	Kali Linux	Dependency installation via native package managers
skills/radare2/scripts/recon.ps1	Windows	PowerShell implementation of Radare2 reconnaissance
skills/radare2/scripts/recon.sh	Kali Linux	Bash implementation of identical reconnaissance logic
docs/ARCHITECTURE.md	Both	System architecture with multi-platform diagram
RULES.md / kali/RULES-kali.md	Platform-specific	AI routing rule sets

Design Principles

The reverse-skill multi-platform script architecture adheres to three core principles evident in the source code:

• Separation of concerns – What the skill does (in skills/) remains independent from how the OS executes it
• Minimal platform wrappers – Only thin adapter scripts differ; no logic duplication
• Bootstrap symmetry – Both platforms implement equivalent dependency resolution despite different package ecosystems

Summary

• reverse-skill separates platform-agnostic skill definitions in skills/ from platform-specific execution scripts
• Windows uses PowerShell modules in skills/scripts/*.ps1 with ToolDiscovery.ps1 for tool resolution
• Kali Linux uses Bash scripts in kali/scripts/*.sh with direct bootstrap-reverse.sh invocation
• Both platforms share identical SKILL.md definitions and output generators
• Platform detection occurs via RULES.md selection before script loading
• The Radare2 recon skill demonstrates functional parity with syntax-appropriate implementations

Frequently Asked Questions

How does reverse-skill detect which platform it's running on?

The system reads the appropriate rules file at initialization. Windows environments use RULES.md while Kali Linux uses kali/RULES-kali.md as specified in docs/ARCHITECTURE.md lines 6-10. This selection determines subsequent script path resolution.

Can I run the same skill on both Windows and Kali without modification?

Yes. The skill definitions in skills/ are completely portable. Only the wrapper scripts differ—you call recon.ps1 on Windows or recon.sh on Kali, but both execute identical analysis steps against the same SKILL.md configuration.

What happens if a required tool like rabin2 is missing?

The bootstrap layer handles installation automatically. On Windows, ToolDiscovery.ps1 checks tool-index.md and invokes bootstrap-reverse.ps1. On Kali, functions like ensure_tool in recon.sh call bootstrap-reverse.sh directly. Both consult their respective bootstrap-manifest.json files to determine installation sources.

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