Repolode
Submit
zinja-coder

zinja-coder/jadx-ai-mcp

A JADX plugin plus Python MCP server that lets LLMs such as Claude analyze decompiled Android APKs live from the JADX GUI.

View on GitHub
Stars
2,993
Forks
281
Open issues
4
Last commit
20 days ago

What is it?

JADX-AI-MCP is a plugin for the JADX Android decompiler that exposes the decompiled app to LLM clients through the Model Context Protocol (MCP). It works as a pair with the separate jadx-mcp-server Python program, which the MCP client launches and which forwards tool calls to the plugin over HTTP.

Through this connection, an LLM can fetch the currently selected class, list methods and fields, read smali, pull the AndroidManifest.xml and string resources, search for classes by keyword, and trace references to classes, methods and fields. It can also rename classes, methods, fields, packages and variables, add comments that are saved with the JADX project, and read stack frames, threads and variables from the JADX debugger.

The project is aimed at security reviewers and reverse engineers who want an assistant for static analysis, such as spotting hardcoded secrets or insecure API calls. It supports stdio and HTTP transports and is configured with clients like Claude Desktop, Cherry Studio and LM Studio.

Key features

  • 01Lets LLM clients fetch the selected class, method, field and smali code from JADX.
  • 02Reads AndroidManifest.xml, manifest components, strings and other resource files.
  • 03Searches classes by keyword and finds cross-references to classes, methods and fields.
  • 04Renames classes, methods, fields, packages and variables, and saves comments with the project.
  • 05Exposes JADX debugger stack frames, threads and variables to the model.
  • 06Supports stdio and HTTP transports with configurable host and port.

Use cases

  • Triaging an Android APK for hardcoded secrets, insecure API usage or risky permissions.
  • Asking an LLM to explain obfuscated classes and suggest readable names during reverse engineering.
  • Walking through execution flow with the JADX debugger and an AI assistant.
  • Analyzing app resources and manifest components during a mobile security assessment.

Quick start

  1. 1

    Download jadx-ai-mcp-<version>.jar and jadx-mcp-server-<version>.zip from the releases page

  2. 2

    Unzip the JADX MCP server archive

    unzip jadx-mcp-server-<version>.zip

Things to know

  • The project is in early development, so bugs, crashes and logic errors are expected.
  • It requires Java 11+ and Python 3.10+, plus JADX and an MCP-capable LLM client such as Claude.

In collections

Reverse engineering toolsMCP serversSecurity scanners, audits & threat intel toolsAndroid apps & phone tools

Topics

#reverse-engineering#android#mcp#llm#security#jadx#decompiler#ai#java#mcp-server

FAQ

What is jadx-ai-mcp?
A JADX plugin plus Python MCP server that lets LLMs such as Claude analyze decompiled Android APKs live from the JADX GUI. JADX-AI-MCP is a plugin for the JADX Android decompiler that exposes the decompiled app to LLM clients through the Model Context Protocol (MCP). It works as a pair with the separate jadx-mcp-server Python program, which the MCP client launches and which forwards tool calls to the plugin over HTTP.
Is jadx-ai-mcp free to use? What license does it have?
jadx-ai-mcp is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is jadx-ai-mcp still maintained?
Actively maintained. The last commit to jadx-ai-mcp was on Oct 19, 2026.
What can I use jadx-ai-mcp for?
Triaging an Android APK for hardcoded secrets, insecure API usage or risky permissions. Asking an LLM to explain obfuscated classes and suggest readable names during reverse engineering. Walking through execution flow with the JADX debugger and an AI assistant. Analyzing app resources and manifest components during a mobile security assessment.
How do I install jadx-ai-mcp?
Download jadx-ai-mcp-<version>.jar and jadx-mcp-server-<version>.zip from the releases page Unzip the JADX MCP server archive: unzip jadx-mcp-server-<version>.zip
Where can I get jadx-ai-mcp?
The source code and installation instructions are on GitHub: https://github.com/zinja-coder/jadx-ai-mcp

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects