> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mira-app.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Starting a Session

> How to create a new chat session, optionally apply a Skill or Workflow, and begin reasoning.

## Creating a new session

Click **New Chat** in the sidebar, or press **⌘N** (macOS) / **Ctrl+N** (Windows/Linux).

A new session is created with an auto-generated title (updated after your first message). The chat window opens empty and ready for input.

## Applying a Skill before you start

On the new chat screen, a **Skill Picker** is available. Skills give MIRA a specialised reasoning persona for the duration of the session — lower temperature for precision-critical domains, higher iteration limits for complex analysis, domain-specific instructions.

Click a skill to select it. The skill name appears in the session header. To remove a skill mid-session, click the skill badge in the header and select **Detach**.

[Browse built-in skills →](/features/skills/built-in-skills)

## Applying a Workflow before you start

If you want a multi-step pipeline rather than a free-form conversation, select a **Workflow** from the Workflow Picker on the new chat screen. MIRA will execute each step in sequence, forwarding outputs from one step to the next.

[Browse built-in workflows →](/features/workflows/built-in-workflows)

<Info>
  A session can have either a Skill or a Workflow active, not both simultaneously. The Workflow
  takes precedence if both are set programmatically.
</Info>

## The chat interface

| Element            | Description                                                                          |
| ------------------ | ------------------------------------------------------------------------------------ |
| **Message input**  | Multi-line text area at the bottom. **Enter** sends; **Shift+Enter** adds a newline. |
| **Send button**    | Alternatively, click the arrow icon.                                                 |
| **Session header** | Shows the active engine, active skill/workflow, and document count.                  |
| **Document Panel** | Collapsible right panel for managing uploaded documents.                             |
| **Status bar**     | Bottom bar showing engine status, context usage, and app version.                    |

## Sending a message

Type your question or instruction in the input area and press **Enter**. The message appears as a user bubble; the assistant's response streams in token by token below it.

While a response is streaming:

* The **Cancel** button (✕) in the input area stops the current query
* Cancelled queries are preserved in the session history so you can see what was generated before cancellation
* Other sessions can still be active — MIRA supports up to 3 concurrent RLM sessions and unlimited NAE sessions

## Session auto-title

After your first message, MIRA sets the session title based on the content of your query. You can rename any session by double-clicking its name in the sidebar.

## Related

<CardGroup cols={2}>
  <Card title="Streaming & Markdown" icon="type" href="/features/chat/streaming-and-markdown">
    How responses are rendered — code blocks, tables, LaTeX
  </Card>

  <Card title="Session Management" icon="list" href="/features/chat/session-management">
    Pin, rename, delete, and manage conversation history
  </Card>
</CardGroup>