💀 JeffreyOS — Someone Built an OS From the Leaked Epstein Files

:skull: JeffreyOS + Torrents + Tools — The Complete Epstein Archive Guide

Someone built a browser-based simulation of Jeffrey Epstein’s personal computer. The documents are real. The aesthetic is unhinged.


3 million pages. 180,000 images. 2,000 videos. The DOJ released everything on January 30, 2026, and the internet did what the internet does — someone turned it into a desktop OS you can explore in your browser. Welcome to digital forensics as performance art.


💻 JeffreyOS — The Project

What it is: A web-based simulation of what Jeffrey Epstein’s personal operating system might have looked like — complete with desktop icons, file browsers, and the actual leaked documents embedded throughout.

Live Demo: jeffreyos.pages.dev

Source Code: github.com/muneebwanee/JeffreyOS

Why This Exists

The DOJ dump was a mess — millions of files thrown into a government portal with terrible navigation, broken redactions, and no organization. JeffreyOS is one person’s answer to that: make the archive explorable instead of just searchable.

Is it tasteful? Absolutely not. Is it effective at making people actually engage with public records? …yes.

📂 The Actual Files — Where They Come From

JeffreyOS pulls from the official releases. Here’s where the documents actually originate:

Source What’s In It Size
DOJ Epstein Library 3+ million pages, 180k images, 2k videos (Jan 2026 release) ~300+ GB across 12 datasets
House Oversight Release ~33,000 pages of emails and estate records (Nov 2025) ~20 GB
Maxwell Case Documents Court filings, depositions, unsealed records Various

Official DOJ Portal: justice.gov/epstein

House Oversight Google Drive: Original release folder

The files include flight logs, contact books, text messages, FBI tip line submissions, BOP video footage, Maxwell proffer session audio, and… a lot of commercial pornography from Epstein’s devices (redacted, but still present in the archive).

🛠️ Better Tools — If JeffreyOS Is Too Spicy

For serious research (or if you just want to search without the desktop cosplay), these tools exist:

Tool What It Does Link
Epstein Docs Archive AI-OCR’d, searchable database of 8,175+ documents with summaries epstein-docs.github.io
Epstein Doc Explorer Graph-based relationship network — see who connects to who github.com/maxandrews/Epstein-doc-explorer
Epstein Text Messages Color-highlighted, human-readable format of his actual texts github.com/michelcrypt4d4mus/epstein_text_messages
FULL_EPSTEIN_INDEX Unified archive combining all releases (House, DOJ, FBI, CBP) github.com/theelderemo/FULL_EPSTEIN_INDEX
Epstein Files Downloader CLI tool to archive DOJ files before they disappear github.com/Surebob/epstein-files-downloader

Why These Matter

The DOJ portal is slow, rate-limited, and files have already disappeared (Datasets 9-11 ZIPs were removed). Community archives are the actual preservation layer.

🧲 Torrent Mirrors — The Files That Won't Die

DOJ links break. Government servers throttle. But torrents are forever (as long as people seed).

Dataset Size Magnet
Dataset 10 82 GB (Full, verified) magnet:?xt=urn:btih:d509cc4ca1a415a9ba3b6cb920f67c44aed7fe1f
Dataset 11 27 GB magnet:?xt=urn:btih:59975667f8bdd5baf9945b0e2db8a57d52d32957
Dataset 12 114 MB magnet:?xt=urn:btih:8bc781c7259f4b82406cd2175a1d5e9c3b6bfc90
Dataset 9 46 GB (partial) magnet:?xt=urn:btih:0a3d4b84a77bd982c9c2761f40944402b94f9c64

Torrent index repo: github.com/yung-megafone/Epstein-Files

Note: Seed if you download. Preservation only works if the swarm stays alive.

⚠️ What's Actually In The Files

A breakdown of what the January 2026 DOJ release contains:

Category Examples
Flight logs Private jet manifests, passenger lists, destinations
Contact books Names, phone numbers, addresses (many redacted, some not)
Emails Epstein’s personal correspondence, assistant communications
Text messages iMessage logs from his devices
FBI tip lines Public submissions to the FBI (unverified allegations)
Court filings Depositions, motions, sealed documents now unsealed
BOP footage Video from the jail where he died
Maxwell proffer Audio recordings of her cooperation sessions
Pornography Seized from devices, females redacted, still present

The Redaction Disasters

The DOJ’s redaction process was… not great:

  • At least 43 victims’ full names were exposed, including more than two dozen who were minors when abused
  • Some names appeared over 100 times, with home addresses visible in keyword searches
  • Unredacted nude images showing young women with faces visible were published before being removed
  • Attorneys who provided a list of 350 victims said the DOJ failed to perform basic keyword searches

This is public record, but it’s also a data protection trainwreck.

🔍 How to Search Effectively

Millions of pages require strategy:

1. Start With Searchable Archives

Don’t raw-dog the DOJ portal. Use:

  • epstein-docs.github.io — OCR’d and indexed
  • Ctrl+F works, entity search works, date filters work

2. Use the Relationship Graph

The Epstein Doc Explorer shows who appears in documents together. Find a name → see their network → find more documents.

3. Download Locally for Deep Dives

Use the epstein-files-downloader CLI:

pip install -r requirements.txt
pip install -e .
epstein-dl download --all --output ~/epstein-archive

Then grep to your heart’s content.

4. Check the Text Messages

The epstein_text_messages repo has already parsed and formatted his texts. Color-coded, human-readable, context notes included.

📊 The Numbers

As of the January 30, 2026 release:

Metric Count
Total pages 3+ million (DOJ claims ~6 million may qualify)
Images 180,000+
Videos 2,000+
Datasets 12
Attorneys/reviewers involved 500+
Storage required 300+ GB
Documents in searchable archive 8,175+ (community OCR’d)

The DOJ’s “collect everything” strategy means this includes verified evidence AND unverified FBI tip line submissions. Not everything in here is true — some of it is just what random people reported.


The files are public. The desktop simulation is optional. The rabbit hole is bottomless.

8 Likes