Communications: Email, Text Message and Phone Research Tools

Jmail presents released emails in a webmail style interface. The public inbox includes message search and views for different source accounts.

1. Search a neutral subject term, such as schedule.
2. Open a relevant message and inspect its actual sender and recipient headers. Distinguish these from names mentioned in the body.
3. Check which account and release the message came from, then confirm the date on the message.
4. Open the message, read the surrounding conversation and follow its document reference.

The familiar inbox design is a reconstruction from released material. It does not establish that every message or attachment survives. Separate the author of a message from people quoted or mentioned inside it.

Access: public inbox without an account prompt. Credit: Jmail, created by Riley Walz and Luke Igel. See EpsteinWiki’s Jmail guide.

Main caution: Identify the release and account. A reconstructed inbox has gaps.

The Yahoo view limits the inbox to the project’s Yahoo collection. Its page credits the rollout to Drop Site News. Jmail’s source documentation distinguishes this collection from House Oversight and DOJ releases.

1. Open Yahoo and confirm the account filter remains selected.
2. Try the same subject term used in the main inbox, such as schedule.
3. Compare dates and full message text across any apparent copies.
4. Record the collection with the citation.

Different releases may contain different versions, redactions or attachments. Do not assume a message is a DOJ record merely because it appears beside DOJ material in the same service.

Access: public inbox view. Credit: Jmail and the source projects identified in its documentation. See EpsteinWiki’s Yahoo collection background.

Main caution: Identify the release and account. A reconstructed inbox has gaps.

JMessage arranges recovered iMessage material as conversations. The current list visibly includes uncertain labels beginning with “Maybe”. Keep that uncertainty when taking notes.

1. Use the sidebar Search field to locate a conversation. Try a public participant already named in your source, such as Joi Ito.
2. Open the conversation and read messages before and after the passage of interest.
3. Record its displayed date and use View source document to inspect the referenced record. Data attribution explains the extraction and the separate basis for name attribution.
4. Compare sender labels, time information and text with the original extraction.

Conversation reconstruction can depend on OCR, ordering and identity matching. An inferred label is not independent identification. Do not infer who a redacted participant must be.

Access: public conversation list and individual conversation pages. Credit: the Jmail project; EpsteinWiki’s JMessage guide credits Jinglin Li and the underlying Cryptadamus text processing work.

Main caution: Preserve uncertain participant labels and verify the source extraction.

Communications lists email records, participants, paired correspondents and resolved identities. Its Emails table exposes Add filter, Apply, sorting and CSV or JSON export.

1. Open Emails and add a filter on the field relevant to your question.
2. For example, use the subject field to look for schedule. Start broadly if the extraction is inconsistent.
3. Inspect the EFTA identifier, page, source, raw date and parsing confidence.
4. Open the underlying document before relying on a name or subject field.

Some displayed rows visibly contain OCR fragments or text in the wrong field. This collection includes communications about investigations and litigation as well as personal correspondence. A row count is not a count of unique messages sent by Epstein.

Access: public catalogue and email table. Credit: Epstein Data. Its identity confidence and match method fields should be preserved when using resolved names.

Main caution: Parsed fields and identity matches can be wrong.

Phone Records organizes extracted billing records, with year, month, time of day and owner filters plus source document links.

1. Start with a period, such as 2005, rather than an unidentified number.
2. Narrow the month or owner if the research question requires it.
3. Inspect the source document list and open the record supporting the relevant entry.
4. Check whether the row describes an attempted call, duration or another billing event.

The page describes uneven coverage across accounts and years. Billing metadata cannot establish the words spoken or who physically used a phone. Do not publish private phone numbers or resolve unnamed contacts through guesswork.

Access: public page and filters. Credit: Tommy Carstensen. The archive’s Sources hub groups the communications material.

Main caution: Call metadata, contact lists and message contents are different evidence.

The Jmail Data API documents public Parquet and NDJSON datasets. It states that these static downloads require no API key or authentication. Its email data includes multiple release sources, so record the batch and collection used.

1. Read the dataset definitions before downloading.
2. Choose full email data when the body is needed. The slim dataset omits bodies.
3. Consult the email client reference for the documented schema and loading method.
4. Begin with a bounded question, such as messages whose subject contains meeting during one month.
5. Keep source identifiers and inspect the original records behind any result.

Downloading a file does not make normalized names, dates or counts error free. Plan for duplicates and forwarded text. Avoid republishing private contact details or sensitive material in a derived dataset.

Credit: Jmail. Its documentation distinguishes the House Oversight, DOJ and Yahoo source collections.

The Cryptadamus repository documents color highlighted email and text views, raw text comparison and tools for finding strings across a local collection.

1. Read the README and follow its link to the generated views.
2. Choose the relevant document or conversation rather than relying only on a curated excerpt.
3. Compare highlighted text with the raw source and the original document.

A useful document lookup example from the project is HOUSE_OVERSIGHT_030999. Highlighting helps navigation but cannot authenticate an attribution. The repository also describes optional local software; running it requires separate setup and source files.

Access: the repository is public without a GitHub account. The large hosted message view could not be fully checked in this review, so no claim is made that all its interactive controls work. Credit: Michel de Cryptadamus. The repository identifies the project’s license and generated views.

Email Metadata is a separate catalogue containing several versions of email, pair, monthly and sender tables. It can help locate candidate time periods or correspondent spellings before reading message content.

1. Choose a table whose fields match the question.
2. For a chronology, look for raw and normalized dates. For correspondence, distinguish sender and recipient fields.
3. Record the exact table version and source identifiers used.
4. Verify a sample of records in the original documents before describing any pattern.

At this review the catalogue was public, but its Emails link redirected to the source repository. Use document search if interactive table access is unavailable. Search schedule with the Email category as a practical fallback.

Do not add counts from different table versions together. Merged records, duplicates, promotional messages and parsing rules can change the denominator.

Credit: Epstein Data. Project data and documentation explain the available exports.

DocETL’s explorer analyzes a defined set of 2,322 emails from the November 2025 House Oversight release. It provides attribute based filtering across fields such as participants, organizations, subjects and email content.

1. Choose the field that matches the question. A subject search for meeting is a straightforward starting point.
2. Open a matching email and read its full text.
3. Treat extracted topics, tone, summaries and concern flags as automated annotations.
4. Verify the passage in the linked original release.

A filter that finds an automated concern label has found an annotation to examine. It has not established wrongdoing. This is a bounded collection, so an absent result cannot settle what exists elsewhere.

Access: public showcase and dataset description; the embedded explorer may require JavaScript. Credit: the DocETL project and the original House Oversight release. See EpsteinWiki’s DocETL guide.

Availability check, October 11, 2026: the email explorer returned “Failed to load emails: Failed to fetch: 500” twice in our browser check. If it does not load, use DocETL’s linked original House Oversight release for source context; that page contains three selected emails.

The Wolff and Epstein archive presents a focused correspondence collection with message search and speaker controls. The page credits Ellie Leonard’s compilation of DOJ released records.

1. Search a neutral word, such as meeting.
2. Use the speaker control to understand which side of the exchange you are reading.
3. Read adjacent messages and check the original document behind the excerpt.

Keep the compiler’s selection and the original release distinct. This view is intended for one correspondence set; it is not a general search of every correspondent.

Access: public page; messages load dynamically. Credit: Epstein Data and Ellie Leonard, whose source compilation is linked on the page. See EpsteinWiki’s existing sleuth report.

Carstensen’s Skype page brings together contacts, calls and messages extracted from seized devices, with source identifiers.

1. Read the source and date range first.
2. Use the page’s sections or browser Find for a date, such as 2016, or a known source identifier.
3. Open the cited extraction and compare the surrounding entries.
4. Keep contact records, call events and message contents separate in your notes.

A contact entry does not prove a conversation occurred. Device copies and redacted handles can complicate both chronology and attribution. Preserve uncertainty instead of merging similar names.

Access: public source linked page. Credit: Tommy Carstensen.

Main caution: Call metadata, contact lists and message contents are different evidence.

The FedEx Courier Timeline can supply a dated physical correspondence lead.

1. Look for an office, city or date; try Zorro Ranch.
2. Open the row’s source receipt.
3. Check whether sender, recipient and delivery information are legible.

Keep package contents and the recipient’s knowledge unresolved unless another record establishes them. Credit: Tommy Carstensen. See EpsteinWiki’s courier guide.

In Epstein Data search, try “meeting” with the Email category, then adjust the source and date controls. You can also search a known EFTA identifier directly. Its content filters can help locate email metadata and attachments. Check the source page rather than assuming a classified result is complete.

Both search interfaces are public. A zero result may reflect spelling, OCR, redactions, filters or coverage. Record the query and filters so the search can be repeated.

Main caution: Read the cited page and its neighboring pages before quoting.

In Carstensen’s corpus search, try “meeting” AND “schedule”. Use the citation resolver when you already have a production stamp. Different OCR and collection boundaries can produce different matches.

Both search interfaces are public. A zero result may reflect spelling, OCR, redactions, filters or coverage. Record the query and filters so the search can be repeated.

Main caution: Read the cited page and its neighboring pages before quoting.

Epstein Web Tracker now describes its main viewer as a Bluesky backed feed. Its former hosted file search has been retired.

1. Choose Search and enter a known document identifier or subject keyword.
2. Open a relevant post and follow its cited record.
3. Verify the record in a document archive before adding it to a research conclusion.

Recent posts, comments and network maps can suggest where to look. They are not an independent archive of every released email. Public reading was available; the page also offers account functions for other uses.