Kali-MCP-Server-Walkthrough

The system is critically vulnerable to Remote Code Execution (RCE) on both the Kali VM and the client machine. The 'kali_server.py' uses 'subprocess.run(command, shell=True)' to execute commands received from the client, with no apparent input sanitization or command whitelisting. This allows an LLM to execute arbitrary system commands on the Kali machine. Furthermore, 'mcp_server.py' on the client side contains 'eval(tool_code)', which can execute arbitrary Python code if 'tool_code' originates from an untrusted source, such as an LLM's output. The design inherently trusts all LLM outputs for command execution, making it extremely dangerous and prone to compromise.