The pool is a rolling sample of posts collected from the Bluesky firehose - the real-time stream of all public posts on the network. It captures original posts (no replies or reposts) to create a diverse dataset for exploring content, training recommendation algorithms, and understanding engagement patterns. Posts are periodically ingested and older ones are pruned to maintain a manageable size.
Posts are collected hourly via systemd timer (~100 posts per run).
Maximum capacity: 100,000 posts (oldest posts are automatically removed when limit is exceeded).
Timer: systemctl status hyperblues-ingest.timer
Manual run: sudo systemctl start hyperblues-ingest.service