Introduction

Turn public web pages into clean data for applications and AI agents. DataBlue handles retrieval, rendering, cleaning, and delivery.

Web Data Built for LLMs

DataBlue removes page noise and returns LLM-ready markdown, JSON, links, images, HTML, or screenshots.

DataBlue is the web-data layer between a source and your model, database, or application.

Five Modes, One Platform

Choose the smallest workflow that produces the result you need.

A simple selection rule

• Known page: use Scrape.
• Known website, unknown page: use Map, then scrape the URLs you select.
• Connected website section: use Crawl with explicit limits.
• Known question, unknown website: use Search.
• Known fields: use Extract.
• Known supported data surface: use its specialized Data API.

Specialized Data APIs

Use a Data API when you need structured records from a supported source instead of general page content.

Use core modes for flexible web retrieval. Use Data APIs for source-specific records. Check Data API Status before production use.

What DataBlue Handles

DataBlue manages the retrieval work behind each request and returns one consistent response shape.

What your application still controls

Performance Without Made-Up Numbers

Timing varies by source, output, and page count. Scrape returns directly; longer work uses jobs.

Benchmark the real sources and formats your product will use.

curl --fail-with-body --silent --show-error \
  -X POST "https://api.datablue.dev/v1/scrape" \
  -H "Authorization: Bearer $DATABLUE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://example.com",
    "formats": ["markdown", "links", "images"]
  }' | jq '{
    success,
    time_taken: .data.time_taken,
    words: .data.metadata.word_count,
    links: (.data.links | length),
    images: (.data.images | length)
  }'

A Basic Request

Export an API key, scrape one stable page, and inspect the response.

export DATABLUE_API_KEY="wh_your_api_key"

curl --fail-with-body --silent --show-error \
  -X POST "https://api.datablue.dev/v1/scrape" \
  -H "Authorization: Bearer $DATABLUE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://example.com",
    "formats": ["markdown", "links", "images", "headings"]
  }' | tee datablue-response.json

{
  "success": true,
  "data": {
    "markdown": "# Example Domain

This domain is for use in illustrative examples.",
    "links": ["https://www.iana.org/domains/example"],
    "images": [],
    "headings": [{"level": 1, "text": "Example Domain"}]
  }
}

If you used the cURL example, inspect the saved response without printing the entire payload:

jq '{
  success,
  source: .data.metadata.source_url,
  title: .data.metadata.title,
  word_count: .data.metadata.word_count,
  markdown_preview: (.data.markdown | .[0:300]),
  link_count: (.data.links | length),
  image_count: (.data.images | length),
  time_taken: .data.time_taken
}' datablue-response.json

A successful first request proves four things

1. Your API key is accepted.
2. DataBlue can retrieve and process the source.
3. The response contains the output formats you requested.
4. Your application can parse the response without depending on undocumented internals.

Use DataBlue From Your Preferred Surface

Continue From Here