wtf is godzilla-ai-mcp?

This repository is a plugin that connects the Godzilla webshell management tool to the Model Context Protocol, or MCP, a standard that lets AI assistants call external tools. Godzilla is a well known tool in the security research and penetration testing community used to manage encrypted connections to webshells, which are scripts placed on a server that let someone control it remotely. The readme states the plugin supports Godzilla versions 4.0.1 and 415, and once loaded, it can be deployed on an internal network for shared, multi-person use. According to its description, the plugin groups its capabilities into several categories. Connection management covers creating, saving, listing, testing, and closing connections to a target system. File operations let a user browse, read, write, upload, download, and delete files on a connected system. There is also support for running system commands and executing code in whatever language a particular webshell uses, such as Java, PHP, C#, or ASP. Additional categories described in the readme include gathering system information such as the operating system, running processes, and network details, taking screenshots of the remote desktop, probing internal databases and running SQL queries against them, injecting and managing so called memory-resident shells inside a Java application server, loading shellcode, and setting up a SOCKS5 proxy or scanning ports through the compromised host to reach other systems on the same network. This is a specialized security tool intended for people who already understand webshell-based access and penetration testing workflows, since Godzilla itself is typically used in authorized security assessments or, in less legitimate cases, unauthorized intrusion. The readme includes screenshots of the plugin's interface but no explanation of who should use it or under what legal or ethical conditions, so anyone evaluating this tool should bring their own judgment about appropriate and lawful use.