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

# Introduction

> The open layer for attention intelligence.

OnlookAI is an open-source SDK and scoring engine that tracks attention signals, quantifies engagement, and triggers interventions — across every surface your users touch.

## What is attention intelligence?

Traditional analytics tell you what users clicked. OnlookAI tells you whether users were actually paying attention — measuring focus depth, scroll behavior, dwell time, idle patterns, and face presence to produce a single interpretable score per session.

## Core primitives

<CardGroup cols={2}>
  <Card title="Track" icon="eye" href="/quickstart">
    Capture attention signals with a single SDK call. Works in browser, native, and embedded environments.
  </Card>

  <Card title="Score" icon="chart-line" href="/concepts/attention-scoring">
    Real-time 0–100 attention score per session. Configurable thresholds, no ML ops required.
  </Card>

  <Card title="Intervene" icon="bolt" href="/concepts/interventions">
    Trigger policy actions, nudges, or alerts when attention drops below threshold.
  </Card>

  <Card title="Audit" icon="shield" href="/concepts/audit-trails">
    Immutable, cryptographically signed audit trails for compliance environments.
  </Card>
</CardGroup>

## Open source

OnlookAI is Apache 2.0 licensed. Fork it, self-host it, extend it.

```bash theme={null}
npm install @onlook/core
```

For compliance and enterprise features, see [Attune Compliance](/enterprise/attune-compliance).