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

# Niblet documentation

> Set up your display, build 64×32 apps locally, and manage Niblet with the CLI or an AI assistant.

Niblet is an app-driven 64×32 ambient display. Use these guides to bring a display online, create local Starlark apps, connect the remote MCP server, and submit immutable releases for review.

<CardGroup cols={2}>
  <Card title="Set up a display" icon="display" href="/getting-started/setup-display">
    Install firmware, connect Wi-Fi locally, and finish the first pixel test.
  </Card>

  <Card title="Build your first app" icon="code" href="/quickstart">
    Install the CLI, create an app, preview it, and upload a review artifact.
  </Card>

  <Card title="Connect Niblet MCP" icon="message-bot" href="/mcp/connect">
    Manage displays and developer app records from Codex, ChatGPT, Claude, Cursor, or another MCP client.
  </Card>

  <Card title="Runtime reference" icon="brackets-curly" href="/reference/widgets">
    Find the widgets, modules, schemas, limits, and compatibility notes used by Niblet apps.
  </Card>
</CardGroup>

## Local-first by design

App source, validation, live preview, and rendering stay on your computer. The developer platform receives source only when you explicitly run `niblet apps
push`, and stores that source with the exact locally rendered WebP as one immutable review artifact.

The remote MCP server manages display state and developer records. It does not receive Wi-Fi passwords, app credentials, source edits, or local render jobs.

<Info>
  Niblet preserves useful Pixlet compatibility. Existing Starlark apps can often move with small metadata or service-integration changes, but compatibility does not waive API terms, asset rights, or physical-device testing.
</Info>