Jadx AI MCP

What zinja-coder/jadx-ai-mcp MCP server does

The zinja-coder/jadx-ai-mcp MCP server links an LLM client with JADX, an Android decompiler, for interactive analysis of APKs. It is delivered as two cooperating parts: a JADX plugin and the separate JADX MCP Server Python component. Together, they give an AI assistant access to the currently decompiled Android application rather than limiting analysis to static text pasted into a conversation.

The project is intended for reverse engineering and static code analysis workflows. Documented examples include quick APK analysis, vulnerability-focused investigation, resource analysis, code review, and assistance while debugging an APK in JADX. It is listed in the Developer Tools category.

How it works

An MCP client sends a tool request to the JADX MCP Server. That server forwards the request to the plugin through HTTP. Request handlers inside the plugin then ask the JADX graphical interface to perform an action or collect information. The result travels back through the plugin and server as an MCP tool result for the LLM client.

This arrangement keeps JADX as the source of the decompiled application context. The model can therefore work with information gathered from the open JADX session, including code and application resources exposed through the integration.

Setup and configuration

The repository combines the JADX-side plugin with the companion JADX MCP Server project. The README points to the project documentation at https://jadx-ai-mcp.readthedocs.io/en/latest/ for current setup instructions. The available project metadata identifies Java 11 or newer and Python 3.10 or newer as the relevant runtime versions.

Those details should be taken from the linked documentation rather than inferred from the repository description.

Tools and capabilities

The integration supports MCP-driven operations around a live JADX session, including:

• Requesting analysis of a decompiled Android APK.
• Gathering code and other context from the JADX GUI.
• Asking JADX to perform supported interface actions.
• Inspecting APK resources through an AI client.
• Reviewing code for possible vulnerabilities.
• Supporting reverse engineering and debugging tasks.

The zinja-coder/jadx-ai-mcp MCP server is useful when an agent needs to combine model reasoning with the current state of a decompiler. The README presents it as a way to move between decompilation, context-aware review, and AI recommendations during one workflow.

Limitations and notes

The project is described as being in an early development stage, so users should expect bugs, crashes, and incorrect logic.

This is not described as a standalone APK decompiler. It depends on JADX and on the companion JADX MCP Server architecture to expose the decompiled application context to an MCP client. The zinja-coder/jadx-ai-mcp MCP server should therefore be evaluated as part of that combined plugin-and-server setup, especially when planning deployment or troubleshooting communication between components.