Agent Zero + NVIDIA NIM: setup and hacking lab demo

This tutorial installs Agent Zero on the presenter's computer and connects it to NVIDIA NIM for model responses. The agent runs locally, while the demonstrated setup uses an external model API. It is not an offline inference walkthrough.

The installation follows the documentation's command line route and selects Quick Start. After setting login credentials, the installer pulls a Docker image. The presenter opens the dashboard at localhost port 5080, then moves to model settings to configure the provider.

The NVIDIA account walkthrough includes email verification and phone verification, which the presenter says is required to obtain an API key. After saving the key in Agent Zero, the presenter selects NVIDIA and the model identified in the transcript as Deep Seek V4 Pro 0813. A short greeting tests the connection. The speaker warns that responses may take time because of demand.

The final demonstration selects the hacker agent preset and asks it to solve a PortSwigger lab without naming the vulnerability. The presenter describes the agent running commands and reports that it solves the lab. This example shows an AI agent workflow for a security exercise; it does not establish reliability across other penetration testing tasks.