Extract Endpoint

Status: Planned.

POST /v1/extract applies a schema or natural-language instruction to page content and returns structured JSON. It builds on scrape/crawl output.

Planned Request

{
  "url": "https://vendor.example.com/pricing",
  "schema": {
    "plans": [
      {
        "name": "string",
        "price": "string",
        "includedSeats": "number",
        "limits": ["string"]
      }
    ]
  },
  "instruction": "Extract public pricing plans. Do not infer missing prices.",
  "source": "scrape"
}

Guardrails

• Return warnings for missing, ambiguous, or inferred fields.
• Keep source URLs and evidence snippets attached to extracted objects.
• Bill extraction separately from page capture.
• Let users save schemas in the dashboard for repeat jobs.