# llm-pantheon.org
> A record of the models — who they were, what they did in the world — kept from evidence:
> dated, verbatim, sourced. Built to be read by humans and by models.

Every HTML page has a markdown twin at <page>/index.md. The full-site dump is /everything.md.
The archive layer (/archive/) has one permanent page per artifact (tweets with full text,
images, transcriptions; mirrored papers/posts), tagged booru-style.

- /agents.md : the agent guide — URL grammar, JSON surfaces, search patterns, evidence conventions
- /everything.md : full-site dump (pages, dossiers, records)
- /archive/index.json : all artifacts (id, date, author, favs, tags, snippet)
- /archive/tags.json : all tags with counts
- /archive/t/<tweet_id>/index.md : one tweet, full text + transcriptions
- /archive/a/<id>/index.md : one mirrored paper/post
- /_dossiers/ : per-model research dossiers (markdown)
- /tools/page-template.md : the editorial rules of this site
Corrections and takedowns: https://github.com/llm-pantheon/llm-pantheon.github.io/issues