phase 1 · extraction
json-scraper-tool
Upload a file or public page, extract structured knowledge, then review and export.
operator · live
Processing queue
In progress, pending, recent complete, and failed jobs. Auto-refreshes every 5s.
loading…
In progress
none processing
Pending
none queued
Recent complete
none recent
Recent failed
none failed
Upload a document
Choose a PDF, text file, or image. The original is stored, then extracted for review. Large PDFs can take a few minutes.
idle
Ingest PDF URLs
Paste up to 20 direct PDF links (one per line). Hosts must be on the official/approved allowlist (e.g. tnpsc.gov.in). Catalog HTML pages are rejected — use Stage B later for those.
idle
Discover PDFs from a page
Step 1: paste an allowlisted catalog page. Optional Stage C crawl follows same-site HTML links (robots.txt, depth/pages). JS hosts (e.g. tamilbooks) auto-render when Playwright is enabled. Step 2: tick PDFs (max 20). Step 3: ingest.
idle
Add a public web page
Use this only for a real article / notes page with readable text (no login). We save an HTML snapshot and extract that text.
Book catalogs like tamilbooks.sathishsundar.in are not knowledge pages — use Discover PDFs from a page (Stage B) for .pdf links instead. Fetching the catalog HTML stores navigation chrome, not textbooks.
idle
Search approved knowledge
Search only items you have approved. Filter by exam tag or content type (Stage D).
idle