All registry data is open (CC BY 4.0). These are the machine-readable surfaces;
each is regenerated from the records on every change.
Formats
- registry.json - every record's structured fields
(normalized taxonomy keys first). For analysis code.
- registry.csv - the same, flat. Loads directly into
pandas or a spreadsheet. Your browser may download rather than display it.
- Per-record markdown - append
.md to any record permalink
(e.g. /pir/2026-0045.md) for the raw source-of-truth record.
- /llms.txt - machine manifest for LLM agents
(llmstxt.org convention) with links to every record's markdown.
- Zenodo archive (DOI) - versioned
snapshots of the entire registry; doi.org content negotiation serves APA/Chicago/BibTeX/CSL
citations from it.
- GitHub repository - full history,
schema, and the verification pipeline.
Statistics caveat: counts are a floor, not a frequency estimate - see
About for the completeness disclosure.