# Chat

> For the complete documentation index, see [llms.txt](https://www.jetbrains.com/help/air/llms.txt).

Use Chat to create and continue tasks. This tool is always available in the workspace: you describe your goal, add the context the agent needs, send the task, and refine the result with follow-up messages.

Chat is also where a finished task reports back. It shows how many lines the task added and deleted, for example +512 -4, and you click the statistics to open the diff.

The indicator next to the chat input shows how much of the model's [context window](context-window.html) the task has used so far.

## Chat toolbar

The toolbar sets how the current task runs. Every option can be changed per task.

* Add context attaches project items to the task input. See [Task context](task-context.html).

* The agent and model selectors choose who does the work. See [Agents and models](select-agents-and-models.html).

* Thinking Level adjusts the reasoning depth for agents that support it. See [Cloud tasks](tasks.html).

* The permission mode controls how freely the agent acts, including [plan mode](plan-mode.html). See [Permission modes](permission-modes.html).

* The environment selector picks where the task runs. See [Run environments](run-environments.html).

* Send starts the task or sends a follow-up message. While the agent is working, it adds your prompt to the [prompt queue](prompt-queue.html).

## What you can do here

* [Define tasks](define-tasks.html) – write a task the agent can act on, and add the context it needs.

* [Cloud tasks](tasks.html) – run the task, and [keep several running at once](multitasking.html).

* [Review changes](review-changes.html) – open the diff from the change statistics and turn your feedback into the next iteration.