SEC EDGAR 8-K Tracker

8-Ks arrive as SGML because the format predates modern parsing. Pick any recent filing and we'll show you the difference between "we have a folder of filings" and "we have a database of events": classification, officer / date / amount extraction, excerpt selection with character-level provenance β€” all in your browser.

Loading 8-K cache…

How this works

  • A nightly Lambda fetches recent 8-K filings for ~10 curated CIKs from the public data.sec.gov API. The raw SGML lands in S3 + CloudFront.
  • Your browser fetches that cache (cached + gzipped).
  • For each filing we parse the SGML header, extract <ITEMS> to classify by 8-K item code (5.02 Executive Departure, 2.01 Completion of Acquisition, …).
  • Regex entity extractors find officers (name ↔ role), dollar amounts (suffix-expanded), and dates (three format variants) β€” each with span offsets into the flattened body.
  • A small TF-IDF ranker surfaces the 2–3 paragraphs most salient to the filing's item code.
  • Send the output to /api/demos/webhook-echo to see exactly what a real integrator would receive.