The Extract endpoint turns a known web page into a table of structured records. Give it a seed URL plus a natural-language description of the rows you want, and it returns one JSON object per line (NDJSON), handling pagination automatically. It's built for 10s–1000s of records from a single listing page.
This uses the REST API, so it needs LINKUP_API_KEY:
test -n "$LINKUP_API_KEY" || echo "Missing LINKUP_API_KEY"
When to use Extract
Use linkup-extract when... |
Use instead... |
|---|---|
| You have one URL and want many structured rows (team, catalog, jobs, speakers) | — |
| You want one page's content as prose/Markdown | linkup-fetch |
| You need to find information across the web | linkup-search |
| You need synthesis across many sources | linkup-research |
How to call it
Always provide a schema for production use (consistent output, clear required fields). Enable verifyUrls when extracted URLs will be used downstream (adds latency).
curl -sS -X POST "https://api.linkup.so/v1/extract" \
-H "Authorization: Bearer $LINKUP_API_KEY" -H "Content-Type: application/json" \
-d '{
"q": "All pricing plans with plan name, monthly price, annual price, features, and usage limits",
"url": "https://competitor.com/pricing",
"schema": {"type":"object","properties":{"planName":{"type":"string"},"monthlyPrice":{"type":"string"},"annualPrice":{"type":"string"},"features":{"type":"array","items":{"type":"string"}}}}
}'
Async lifecycle
POST /v1/extract returns {id, status: "pending"}. Poll GET /v1/extract/{id} about every 30s (crawls run longer than research). When completed, the output has a resultUrl (24h expiry) — download it and parse one JSON object per line.
If you don't know the URL yet, first discover it with linkup-search ("Find the careers page URL for {company}"), then extract from the result.
For the full parameter reference, patterns, and pricing notes, read references/LINKUP_SPECIALIZED_ENDPOINTS.md (Extract section) in this skill's directory.