Reference Layer · Public · Free · No AI

Epstein File Reference

Full-text keyword search across 2,368 documents and ~30,000 pages from the DOJ Epstein release. Every hit opens a page preview — extracted text + the page JPG, no PDF download required. Every page is its own URL.

Sister to the conversational Oracle · all source material mirrored at R2

Tips: phrases auto-quote when separated by spaces · use AND / OR / NOT for boolean · use * as wildcard (e.g. maxw*) · click ↗ to open the PDF jumped to that page

Every hit links to the original PDF on Cloudflare R2 with a #page=N anchor that browsers honor to jump to the matched page. Text was extracted from the source DOJ PDFs (no OCR yet for the 153 scan-only docs). For interpretive questions across the email corpus, use the Oracle instead — this is the dumb-search layer, built for researchers who want primary sources without commentary.