Data and API

Endpoints

What	Where
Index of all records	api/v1/index.json
Every record, in full	api/v1/bundle.json
One record	api/v1/<type>/<ID>.json, e.g. api/v1/mechanism/M-0001.json
JSON Schemas for each record type	api/v1/schema.json
CSV per type	downloads/mechanism.csv, implementation.csv, claim.csv, concept.csv, organization.csv, source.csv
LLM-friendly index	llms.txt
LLM-friendly full text	llms-full.txt

Reading the data

• IDs are stable and never reused. The prefixes are:
  • M: mechanism
  • I: implementation
  • C: claim
  • K: concept
  • O: organization
  • S: source
• Facts and assessments are separated. Readiness assessments, flaws and editors' syntheses carry "assessment": true. Each readiness assessment also carries rubric_version, confidence and status. Treat assessments as editorial judgments, not facts.
• Sources may carry organizations: the IDs of the organizations that authored or published them.
• flags_all includes computed flags:
  • ai-drafted: not yet reviewed by a human expert;
  • review-overdue: past its review interval;
  • critical-flaw-open: has an open flaw of critical severity.
• Versions. schema_version versions the data format. Content changes are tracked in the repository's history.

License

Content is licensed CC BY 4.0. Please cite the record URL and the date you accessed it.