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

# Manual

> Doppler SDK manual integration

Once you have created an account on [Doppler](https://askdoppler.com) and created an API Key, you can proceed with the next steps.

## Installation

Pick your package manager and install the core SDK.

<CodeGroup>
  ```json npm theme={null}
  npm install @askdoppler/core
  ```

  ```json yarn theme={null}
  yarn add @askdoppler/core
  ```

  ```json pnpm theme={null}
  pnpm add @askdoppler/core
  ```

  ```json bun theme={null}
  bun add @askdoppler/core
  ```
</CodeGroup>

## Setup

Make sure to define the following environment variable to store your API Key.

```bash theme={null}
DOPPLER_API_KEY=your-api-key
```

## Usage

The core SDK exposes 2 methods:

* `getSource`
* `logCrawl`

The SDK is designed to sit in a middleware layer and be called on every request, then passing the URL & headers to the `getSource` function and based on the result, you can call the `logCrawl` function to log the traffic data to Doppler's API.
Ideally you don't await logCrawl in order not to slow down your request.

### getSource

The `getSource(headers: Record<string, string>, url: string)` function analyzes incoming requests to detect the traffic source. It accepts a request object with headers and URL, and returns a `DopplerDetectionResult` object.

#### Parameters

#### Return type

```typescript theme={null}
interface DopplerDetectionResult {
  source: string | null; // The detected source (e.g., 'openai', 'perplexity', 'google', 'bing')
  intent: string | null; // The user's intent ('browse' or 'search')
  type: string | null; // The type of detection ('click' or 'crawl')
  highlightedText: string | null; // Any highlighted text from the URL fragment
  detected: boolean; // Whether a source was detected
}
```

### logCrawl

The `logCrawl(payload: DopplerCrawlPayload)` function sends traffic data to Doppler's API. It accepts a payload object and an optional API key.

#### Parameters

```typescript theme={null}
interface DopplerCrawlPayload {
  url: string; // The URL being crawled
  source: string; // The traffic source
  intent: string; // The user's intent
  type: string; // The type of detection
  highlightedText?: string; // Optional highlighted text
  timestamp?: number; // Optional timestamp
}
```

## Test

To test that your setup is correctly working you can try asking [ChatGPT](https://chatgpt.com) or [Perplexity](https://www.perplexity.ai) to search for your websites URL. Data should start showing up right after your request.