Data
Open data
Every record on this site is a row in one of the JSON files below. The files are the source; the pages are rendered from them at build time and nothing on a page exists that is not in a file.
Files
| File | Records | Size | What |
|---|---|---|---|
| being-dossiers.json | 20 | 216.0 KB | 20 categories; 65 source-specific profiles with documentation and visual contracts |
| beings.json | 20 | 232.7 KB | being records (SCHEMA §3) |
| crafts.json | 19 | 124.2 KB | craft records (SCHEMA §3) |
| documents.json | 114 | 437.2 KB | document records (SCHEMA §3) |
| events.json | 80 | 272.2 KB | event records (SCHEMA §3) |
| evidence-reviews.json | 13 | 8.9 KB | |
| footage.json | 18 | 43.3 KB | footage records (SCHEMA §3) |
| map.json | 164 | 548.9 KB | compact points + entity index the map reads |
| media-photo-review.json | 3 | 23.7 KB | |
| media-visual-v2.json | 3 | 4.8 KB | |
| media.json | 194 | 268.5 KB | record galleries; source credits, rights and current image provenance |
| media_additions.json | 4 | 99.6 KB | |
| people.json | 194 | 707.2 KB | person records (SCHEMA §3) |
| sites.json | 71 | 302.3 KB | site records (SCHEMA §3) |
Field definitions, slug registry and conventions: schema.md. Built 2026-09-06 06:30:32 UTC.
Licence
Structured data (the JSON files above, and the record pages rendered from them): Creative Commons Attribution 4.0. Reuse it, correct it, build on it; credit "UAP Record, uap.ab7.ai" and link back.
Transcripts under /corpus/: Machine captions of public videos by American Alchemy (Jesse Michels); rights remain with the creator; provided for research and citation. They are not covered by the CC licence.
Quotes inside records are verbatim from those captions, ≤300 characters, with the stamp of the block they sit in. Caption errors are kept as they are rather than silently corrected.
Change log
Append-only. Every change the curator agent or a human writer makes to the data lands here as one line: when, who, what, on which record, and whether it was applied.
No change log has been written yet.
Research update · 6 September 2026
Visual-follow-up dataset
Six source-led dossiers with media provenance and nine unique supplied post IDs. Media copyrights are separate from the data licence.