JADX-MCP-SERVER is a Python-based MCP server that interfaces with the JADX-AI-MCP plugin to provide live reverse engineering capabilities to local LLMs. It acts as a bridge between the JADX GUI and your AI agent, enabling tool calls such as decompiling code, inspecting AndroidManifest.xml, extracting resources, and querying class/method details in real time. The server exposes a set of MCP tools that can be invoked by the LLM client to retrieve decompiled code, navigation data, and project metadata, which the AI can then analyze and respond to with guidance or automated actions. To use it, run the server and connect your MCP-enabled client (e.g., the JADX-AI-MCP plugin) to the configured host and port. Once connected, you can invoke tools like get_all_classes(), get_method_by_name(), get_android_manifest(), get_strings(), and others to perform live reverse engineering tasks guided by the LLM.

Prerequisites:

• Python 3.10 or newer installed on your system
• Java 11+ installed if you plan to run JADX locally or rely on JADX tooling
• Access to the repository containing jadx-mcp-server

Steps:

1. Clone the repository: git clone https://github.com/zinja-coder/jadx-ai-mcp-server.git cd jadx-ai-mcp-server

2. (Optional) Create and activate a virtual environment: python -m venv venv

   Windows

   venv\Scripts\activate

   macOS/Linux

   source venv/bin/activate

3. Install dependencies (if a requirements.txt is provided): pip install -r requirements.txt

   If no requirements file is present, ensure you have any listed dependencies installed manually

4. Configure environment variables (if needed):

   • HOST: binding address (default 0.0.0.0)
   • PORT: port to listen on (default 5000)

5. Run the MCP server: python -m jadx_mcp_server

6. Verify the server is running by curling the endpoint or checking logs: curl http://localhost:5000/health

7. Ensure the JADX-AI-MCP plugin is configured to point at the server address and port you started the MCP server on.

Notes and tips:

• This server is in an early development stage; you may encounter bugs or incomplete edge cases when parsing large Android projects.
• Ensure the Java/Jadx integration used by the plugin is compatible with your Android APKs and manifests.
• Common environment variables to consider: HOST, PORT, JAVA_HOME (if required by your Java tooling).
• If you experience connection issues, check firewall rules and ensure the host/port are accessible from the client.
• The tool list (e.g., get_all_classes(), get_method_by_name(), get_android_manifest()) represents the available MCP tool calls. Your LLM prompts can reference these tools to drive reverse engineering tasks.
• When updating the server, re-run the server process to pick up any code changes or dependency updates.