these are standard atproto endpoints. you can look at the atproto api reference for more info.
the following are implemented currently:
com.atproto.repo.getRecord(also seeblue.microcosm.repo.getRecordByUri)com.atproto.repo.listRecordscom.atproto.repo.describeRepo(also seesystems.gaze.hydrant.describeRepo)com.atproto.sync.getRepo(sinceparameter not implemented!)com.atproto.sync.getHostStatuscom.atproto.sync.listHostscom.atproto.sync.getRepoStatuscom.atproto.sync.listReposcom.atproto.sync.getLatestCommitcom.atproto.sync.requestCrawl(adds the host to firehose sources)com.atproto.sync.subscribeRepos(WebSocket firehose stream, requiresrelayfeature; slow consumers may receive aConsumerTooSlowerror frame before the connection closes, and acursorpast the last seq gets aFutureCursorerror frame and a close, as the spec says)
the record endpoints (getRecord, getRecordByUri, listRecords, both describeRepos, systems.gaze.hydrant.countRecords), getRepo and getLatestCommit don't serve a repo whose account is takendown, suspended, deactivated or deleted, even while its records are still stored. getRecord and getRecordByUri answer RecordNotFound with the reason in the message, the rest answer with their own error for the status or InvalidRequest.