Docs/Agent mode/Quick start

Agent mode — Quick start

Agent mode lets the AI work through a task on its own: reading your files, creating or editing several of them, running commands, and asking for your approval before anything sensitive.

Turning it on

Select Agent mode

At the bottom of the chat panel, next to the input box, click the mode selector and choose Agent. It is the default mode when Nexora opens.

The other two modes: Chat (plain conversation, no action on your files) and Plan (the AI thinks it through and proposes a course of action without carrying it out yet).

Describe the task

Give it a goal rather than a list of precise instructions — “add pagination to the user list”, “fix the login bug reported in issue #42”. The agent explores your code, plans the steps and carries them out.

Follow the progress

For a multi-step task, a task list appears above the input box and updates as work proceeds. Every file change is shown with a preview of the added and removed lines, and the model’s reasoning is available by expanding the “Thinking” banner.

Approve sensitive actions

Creating a file or running a terminal command always triggers a permission request, shown just above the input box:

AllowApproves this one action, once

Always allowApproves this kind of action for the rest of the session, without asking again

RejectDeclines and lets the agent propose something else

💡Reading files and searching your code never needs permission — only actions that change something (files, terminal) ask for it.

Choosing the model

The model you pick at the top of the chat applies to Agent mode too — see the chat quick start for what each plan includes.

⚠️Agent mode can use more credits than a plain chat, because every file read, search and action is part of the exchange with the model. Your usage stays visible in your dashboard.

← Autocomplete Inline edit →