SimoneAvogadro/android-reverse-engineering-skill

Claude Code skill to support Android app's reverse engineering

What it solves

This project provides a specialized toolset for reverse engineering Android applications to extract their HTTP APIs and understand their internal logic. It simplifies the process of decompiling APK, XAPK, JAR, and AAR files and overcomes common obstacles like R8 obfuscation, which typically renames classes to meaningless strings.

How it works

The toolset integrates several decompilers (jadx, Fernflower/Vineflower) and provides a suite of scripts to automate the analysis workflow:

  • Fingerprinting: Quickly triages an app to detect its framework (e.g., Flutter, React Native), HTTP stack, and obfuscation level before full decompilation.
  • Decompilation: Converts binary files into readable Java/Kotlin code.
  • Kotlin Name Recovery: Uses Kotlin metadata that survives R8 obfuscation to map obfuscated class names (like a.b.c) back to their original names (e.g., LoginRepository).
  • API Extraction: Scans the decompiled code for endpoints, URLs, and authentication patterns across various stacks, including Retrofit, OkHttp, Ktor, Apollo (GraphQL), and Koin.
  • Call Flow Tracing: Allows users to trace the path from UI components (Activities/Fragments) through ViewModels and repositories to the final HTTP call.

Who it’s for

Security researchers, penetration testers, malware analysts, and developers interested in interoperability analysis or educational CTF competitions.

Highlights

  • R8 Deobfuscation: Recovers original Kotlin class names from metadata, making obfuscated code significantly more readable.
  • Broad Stack Support: Extracts APIs from modern Kotlin Multiplatform (KMP) stacks like Ktor and Apollo GraphQL.
  • API-First Workflow: Dedicated scripts for finding API calls and hardcoded URLs.
  • Multi-Engine Support: Supports multiple decompilation engines for side-by-side comparison of output.
  • Claude Code Integration: Can be installed as a skill for the Claude Code CLI tool.

Related

  • Project
    max-sixty/worktrunk

    A CLI for git worktree management designed to let developers run multiple AI agents in parallel by simplifying the creation and switch between separate working directories.

  • Project
    EricLBuehler/mistral.rs

    A fast, flexible LLM inference engine written in Rust that supports multimodal models, agentic tool execution, and OpenAI/Anthropic compatible APIs.

  • Project
    deepseek-ai/DeepGEMM

    A high-performance CUDA kernel library for LLMs that provides optimized GEMM and fused MoE primitives for NVIDIA SM90 and SM100 GPUs.

  • Project
    wang-xinyu/tensorrtx

    A collection of deep learning network implementations using the TensorRT network definition API for flexible, debuggable, and high-performance inference on NVIDIA GPUs.