jddlab Skills

Skills are Markdown knowledge bases that teach AI assistants how to perform specific reverse-engineering tasks using jddlab tools. Each skill lives in its own subdirectory and follows the SKILL.md format used by opencode and Claude Code - the installer can also register them with Codex and VS Code Copilot.

Available Skills

Directory	What it covers
`android-apk-patch/`	End-to-end Android APK patching: decompile → modify smali/resources → recompile → align → sign → install
`java-deobfuscation/`	Detect and reverse Java/Android obfuscation (ProGuard/R8, string encryption, control-flow) with java-deobfuscator & simplify
`ssl-unpinning/`	Bypass TLS certificate pinning: android-unpinner, Frida gadget (apk-patcher), objection, and manual smali patches
`secret-scanning/`	Find hardcoded secrets, keys, and endpoints with APKscan plus manual follow-up
`native-jni-analysis/`	Analyze native `.so`/JNI code: extract JNI bindings and decompile with Ghidra headless

Prerequisites

Skills use jddlab tools - every tool runs inside the jddlab Docker container so nothing needs to be installed locally beyond Docker. Make sure the jddlab MCP server is configured first:

python mcp/install.py add vscode        # VS Code / GitHub Copilot
python mcp/install.py add claude-cli    # Claude Code / Claude CLI
python mcp/install.py add claude-desktop
python mcp/install.py add codex

Installing Skills

Using the jddlab launcher (recommended)

The jddlab / jddlab.cmd launchers include a skills subcommand that works from a repository checkout without any additional setup.

# Install into the current project (default scope)
jddlab skills add claude-cli
jddlab skills add vscode
jddlab skills add opencode
jddlab skills add codex
jddlab skills add all

# Install for all users (user scope)
jddlab skills add claude-cli --scope user
jddlab skills add all --scope user

# Remove
jddlab skills remove claude-cli
jddlab skills remove all

# List available skills
jddlab skills list

On Windows use jddlab.cmd with the same arguments.

Using Python directly (from repo checkout)

# Default scope is 'project' (installs into current directory)
python skills/install.py add opencode
python skills/install.py add claude-cli
python skills/install.py add vscode
python skills/install.py add all

# User scope - available in every project
python skills/install.py add all --scope user

# Remove
python skills/install.py remove opencode
python skills/install.py remove all

Windows shortcuts (pass extra flags after the script name):

skills\install-skills-opencode.cmd
skills\install-skills-claude-cli.cmd
skills\install-skills-vscode.cmd

Adding a New Skill

Create a directory: skills/<skill-name>/
Write skills/<skill-name>/SKILL.md following the existing template.
Run python skills/install.py add all to push it to your AI clients.

File format

Each SKILL.md is a Markdown document that must start with a YAML front-matter block giving the skill's name (matching the directory) and a description of when to use it - Claude Code and opencode use these to discover and trigger the skill:

---
name: my-skill
description: One or two sentences on what this does and when to use it.
---

# My Skill
...

Code blocks use json fences for MCP tool call examples so AI clients can copy them directly.