42.4K ·GitHub
What is this?
It is a skill pack that routes AI coding agents to the right reverse engineering, penetration testing, and security research methodology for a given task (APK, binary, JS, CTF, malware, etc.), checks which tools are installed, and runs a repeatable workflow instead of letting the agent guess commands.
It is a skill pack that routes AI coding agents to the right reverse engineering, penetration testing, and security research methodology for a given task (APK, binary, JS, CTF, malware, etc.), checks which tools are installed, and runs a repeatable workflow instead of letting the agent guess commands.Is this a standalone tool or something I load into my AI client?It is a collection of skill modules and a routing configuration (RULES.md, routing.json, MASTER-ROUTING.md) that an AI coding client loads as project instructions; it is not a separate app you run on its own.
It is a collection of skill modules and a routing configuration (RULES.md, routing.json, MASTER-ROUTING.md) that an AI coding client loads as project instructions; it is not a separate app you run on its own.How do I get it running?Clone/fork the repo into your project, then refresh the tool index for your platform (PowerShell script on Windows, bash script on Linux/macOS/Kali) so it detects which reverse engineering tools you have installed.
Clone/fork the repo into your project, then refresh the tool index for your platform (PowerShell script on Windows, bash script on Linux/macOS/Kali) so it detects which reverse engineering tools you have installed.What do I need installed before using it?You need Java/JDK (for jadx and apktool), Node.js 22.12+ (for JS toolchain and MCP servers), Python 3.x (for Frida and helper scripts), and a compatible AI coding client such as Claude Code, Codex, Cursor, or OpenCode.
You need Java/JDK (for jadx and apktool), Node.js 22.12+ (for JS toolchain and MCP servers), Python 3.x (for Frida and helper scripts), and a compatible AI coding client such as Claude Code, Codex, Cursor, or OpenCode.Which AI clients does it actually support?The routing core is designed to be client-neutral and tested on Windows and Ubuntu CI, with Claude Code, Codex, Cursor, Cline, Kiro and OpenCode all supported; a dedicated adapter plugin currently exists only for Codex.
The routing core is designed to be client-neutral and tested on Windows and Ubuntu CI, with Claude Code, Codex, Cursor, Cline, Kiro and OpenCode all supported; a dedicated adapter plugin currently exists only for Codex.Does it work on Windows, Linux, and macOS?Yes, it ships platform-specific tool-index scripts and docs for Windows, Linux/Ubuntu-Debian, macOS, and Kali Linux, and CI runs the regression tests on both Windows and Ubuntu.
Yes, it ships platform-specific tool-index scripts and docs for Windows, Linux/Ubuntu-Debian, macOS, and Kali Linux, and CI runs the regression tests on both Windows and Ubuntu.What scenarios does the skill matrix cover out of the box?44 routing rules (R0-R45) cover APK/Android, iOS, binary reverse (exe/dll/so/elf), Binary Ninja, .NET, frontend JS, DSL VM opcode reverse, HTTP capture, malware/YARA, pentesting, red-team attack chains, CTF, firmware/IoT, patch diff/N-day, pwn/exploit dev, EDR bypass, API/GraphQL, supply chain, and LLM/AI security.
44 routing rules (R0-R45) cover APK/Android, iOS, binary reverse (exe/dll/so/elf), Binary Ninja, .NET, frontend JS, DSL VM opcode reverse, HTTP capture, malware/YARA, pentesting, red-team attack chains, CTF, firmware/IoT, patch diff/N-day, pwn/exploit dev, EDR bypass, API/GraphQL, supply chain, and LLM/AI security.How does the routing actually decide which tool or playbook to use?A single structured config (skills/config/routing.json) defines the rules; MASTER-ROUTING.md gives a fast triage ladder and skills/routing.md is the full task-to-skill matrix, validated by a 175-case regression benchmark run in CI.
A single structured config (skills/config/routing.json) defines the rules; MASTER-ROUTING.md gives a fast triage ladder and skills/routing.md is the full task-to-skill matrix, validated by a 175-case regression benchmark run in CI.What are the key files I should look at first?READMEAI.md is the AI agent bootstrap, RULES.md sets global routing scope rules, skills/SKILL.md is the master entry point, and skills/tool-index.md shows which tools are detected on your machine.
READMEAI.md is the AI agent bootstrap, RULES.md sets global routing scope rules, skills/SKILL.md is the master entry point, and skills/tool-index.md shows which tools are detected on your machine.Why did the author build a routing layer instead of just writing prompts?Because AI agents don't inherently know whether to use jadx, apktool, Frida, IDA, or BurpSuite for a given task, and experience from past tasks wasn't being reused — the project centralizes that decision logic and an evolving knowledge base.
Because AI agents don't inherently know whether to use jadx, apktool, Frida, IDA, or BurpSuite for a given task, and experience from past tasks wasn't being reused — the project centralizes that decision logic and an evolving knowledge base.What are similar projects?Seep-Reverse-Lab, awesome-reverse-engineering, dsh-redteam-model, x64dbg, Claude-Red, hacker-roadmap, Anthropic-Cybersecurity-Skills, reverify
Seep-Reverse-Lab, awesome-reverse-engineering, dsh-redteam-model, x64dbg, Claude-Red, hacker-roadmap, Anthropic-Cybersecurity-Skills, reverifyWhich topics is it in?malware-analysis, ctf, binary-analysis, powershell, pentesting, penetration-testing, hacking, reverse-engineering, cybersecurity, security-tools, agent-skills, claude-code
malware-analysis, ctf, binary-analysis, powershell, pentesting, penetration-testing, hacking, reverse-engineering, cybersecurity, security-tools, agent-skills, claude-code