# Tweetkeep

> Free public archive of X (Twitter) accounts: tweets, replies, top replies and snapshots, with every deleted or edited tweet flagged and kept.

Every tweet we catch is rechecked after 1 hour, 6 hours, 1 day, 3 days and 1 week. One X says is gone twice, an hour apart, while its account is still up, is marked deleted and kept with its media and page snapshots. Every text, edit and snapshot, and X's raw response for every tweet (WARC), is hashed (SHA-256) and the hashes are timestamped in Bitcoin through OpenTimestamps each hour; a few hundred fresh tweets a day, from the accounts that delete the most, are also saved by the Internet Archive straight from X.

## Pages
- [Home](https://tweetkeep.com/): recently deleted tweets and the accounts we archive
- [Manifesto & FAQ](https://tweetkeep.com/faq): how deletions are detected and how to check our proofs
- [Saved tweets](https://tweetkeep.com/saved): single tweets people asked us to keep, deleted ones recovered from the Internet Archive
- [Stats](https://tweetkeep.com/stats): deletions per day, how long deleted tweets stayed up, which accounts delete the most
- https://tweetkeep.com/tag/<tag>: the accounts with a tag and their latest deleted tweets
- https://tweetkeep.com/@<handle>: an account's archive (deleted and edited tweets first, then the timeline)
- https://tweetkeep.com/p/<tweet id>: a deleted, edited or saved tweet on its own page
- https://tweetkeep.com/deleted.atom and https://tweetkeep.com/@<handle>/deleted.atom: deleted and edited tweets as Atom feeds
- https://tweetkeep.com/oembed?url=https://tweetkeep.com/p/<tweet id>: oEmbed for embedding an archived tweet (an iframe of /p/<id>/embed)
- https://tweetkeep.com/search?q=<words>: search every archived tweet ("exact phrase", -word, or; filters: account, only=deleted|edited, kind=posts|replies, since, until)

## Data (documented at https://tweetkeep.com/api)
- https://tweetkeep.com/@<handle>/archive.json: everything we keep for an account, as JSON
- https://tweetkeep.com/deleted.json: the daily files of deleted and edited tweets; each day at https://tweetkeep.com/deleted/<YYYY-MM-DD>.jsonl (JSON Lines)
- https://tweetkeep.com/search.json?q=<words>: search what archived tweets say, as JSON
- https://tweetkeep.com/proof/<tweet id>: plain-text receipt: the hashes and where they were timestamped
- https://tweetkeep.com/warc/<record id>.warc.gz: X's raw response for a tweet, as a WARC record (linked from its proof receipt)

## Memento (RFC 7089)
- https://tweetkeep.com/timemap/link/<tweet link>: every copy we hold of a tweet (x.com or twitter.com link)
- https://tweetkeep.com/timegate/<tweet link>: redirects to the copy closest to the Accept-Datetime header
- https://tweetkeep.com/web/<YYYYMMDDhhmmss>/<tweet link>: the tweet as we archived it