SiteLens · Targeted Extraction

Find the data you need. Visually or by description.

Click an element or describe what you’re looking for. SiteLens generates reusable extraction instructions for your application. Built for developers. Powered by the SiteLens API.

Try Targeted Extraction View API Documentation Interactive demo

Two ways to identify a field

Visual extraction

Click an element on the webpage preview to identify and extract its content. Hover highlights elements like a browser inspector; a click locks the element so you can review the value and locator.

Description-based extraction

Describe what you want in ordinary language. Example: Professional plan monthly price. SiteLens identifies likely matching elements on the preview and presents ranked candidates for confirmation.

Confirmed selections produce reusable locators for POST /capture. Matching is deterministic—text, labels, and page structure—not AI. Ambiguous descriptions return more than one candidate; you choose. Later capture requests use those locators, not the description.

How it works

Preview → Integrate
Enter URLSelect or describeConfirmTestCopy requestIntegrate

The dashboard Targeted Extraction workspace is an interactive tool for creating and testing API extraction requests. Your application owns the request body, any schedule, and any historical data. SiteLens does not store saved extraction configurations, continuously monitor websites, or track changes over time.

1 · Enter URL

Preview a public page as it appears in a browser, including content loaded by JavaScript. Description matching uses that same preview.

2 · Select or describe

Click an element, or describe the content. Description search returns ranked candidates with scores that are match ranks, not probabilities.

3 · Confirm target

Review the value and recommended locator, name the target, and add it. Nothing is kept as a stored SiteLens configuration.

4 · Test extraction

Send the request to POST /capture. A pass confirms the page as it is now.

5 · Copy API request

Copy the JSON. Later calls use the locators (css, testId, role and name, or text)—not the description.

6 · Integrate

Put the request in your application. Call SiteLens when you need a current value. Scheduling and history stay in your system.

What you can extract

Text & HTML

Headings, paragraphs, labels, and snippets as plain text or markup.

Attributes

Link href values and other attributes when you need URLs or IDs.

Input values

Form field values when the rendered page shows them. Sensitive input values are left out of previews.

Extract What You See

Real browser

Extract content directly from rendered webpages, including elements loaded dynamically by JavaScript. SiteLens works with the page as it appears in a browser, not just its original HTML.

Multiple named targets

One request can extract several fields—starter_price, professional_price, product_availability—each with its own locator and extract kind. Ambiguous locators are flagged so you accept match: "first" deliberately.

Developer API integration

POST /capture

Copy a request for POST /capture. Authenticate with Authorization: Bearer sitelens_sk_… and keep the API key on your backend. Each call returns the current values for the locators you confirmed. 

{
  "url": "https://example.com/pricing",
  "mode": "browser",
  "fallback": "browser",
  "includeExtraction": false,
  "locators": {
    "starter_price": { "css": ".plan-starter .price" },
    "professional_price": { "testId": "pro-price" },
    "availability": { "role": "status", "name": "In stock" }
  },
  "targets": [
    {
      "name": "starter_price",
      "locator": "starter_price",
      "extract": "text",
      "match": "unique"
    },
    {
      "name": "professional_price",
      "locator": "professional_price",
      "extract": "text",
      "match": "unique"
    },
    {
      "name": "availability",
      "locator": "availability",
      "extract": "text",
      "match": "unique"
    }
  ]
}

Successful responses include a targets array with status (matched, missing, ambiguous, or unsupported), value / normalizedValue, and the locator used. A target miss does not fail the overall capture outcome.

Endpoint: POST https://api.sitelensapi.com/capture. Author in the Targeted Extraction workspace, then copy the request.

Real-world use cases

Prices & plans

Pull subscription or product prices into quotes, catalogs, or billing tools.

Availability

Read stock or status copy from public product pages when you need a current snapshot.

Content fields

Extract specific headings, paragraphs, or sections for downstream processing.

Application integration

Call SiteLens from your backend when your application needs a current value. SiteLens does not store history, saved extraction configurations, or monitoring schedules, and it does not send change notifications.

Start selecting

Open the dashboard workspace to create and test a capture request, then copy it into your application.

Try Targeted Extraction View API Documentation Pricing