What it does
The AI Operator lets you give the agent a plain-English objective and it drives the tools to accomplish it. It picks the right tools, runs them, reads the output, and decides what to do next. Everything it runs lands in your task the same as if you had run it manually.
You need an AI provider set up in Settings before the Operator will work. See the AI Providers doc for setup instructions.
Giving it an objective
Open the Agent tab and type what you want done. Be specific about the target and what you want to find. For example: enumerate 10.10.10.10, find open ports and services, and look for exploitable vulnerabilities. The more specific you are, the better the results.
The agent works through the objective step by step. You can watch it run in the Agent tab and see the commands it executes in real time.
Command approvals
You can require the agent to pause before each command and ask for your approval. When approval mode is on, it shows you the exact command it plans to run and waits for you to confirm. This is recommended for sensitive environments or when your rules of engagement require you to review every action. See Command Approvals & Guardrails for more detail.
Stealth level
Use the stealth dropdown in the top right to control how aggressively the agent scans. Lower stealth means fewer probes and less network noise. Set this before starting the agent if your engagement requires low-impact testing.
Chat tab
The Chat tab lets you ask follow-up questions about anything in the task without giving the agent a new objective. Use it to ask the agent to explain a finding, clarify a piece of output, or suggest a next step, without triggering a new autonomous run.
Switching models
You can switch AI models on the fly from the Agent, Chat, and tool output panels. If you want a faster model for quick questions and a more capable one for complex tasks, you can change it without leaving the app.