Scrape a page - url2llm

The draft POST /v1/scrape request requires a URL and at least one enabled format.

{
  "url": "https://example.com",
  "formats": {
    "markdown": true,
    "screenshot": true
  },
  "screenshotParams": {
    "fullPage": true
  }
}

Omit screenshotParams for a viewport screenshot. Screenshots are JPEG, with optional fullPage and quality settings. Quality defaults to 80. Custom viewport dimensions belong in sharedParams.viewport. For language or country preferences, add shared settings:

{
  "url": "https://example.com",
  "formats": {"markdown": true},
  "sharedParams": {
    "languages": ["de-DE", "en-US"],
    "country": "DE"
  }
}

Language and country are independent. A language preference does not require country routing. Country selection depends on supported network egress and does not change an explicit language tag. maxAgeMs defaults to 0 and storeInCache defaults to false. Requests with headers, actions, or profiles bypass document caching. Zero data retention is a separate capability. The deadline defaults to 30 seconds. Atomic completion is the default: all requested outputs must succeed. The proposed return-partial mode uses output errors and an explicit partial flag; its implementation and billing policy remain pending. Authenticate with Authorization: Bearer YOUR_API_KEY when this contract is implemented. Keep target-site headers in sharedParams.headers, separate from your url2llm API key.