Skip to content

Developers & AI agents

The whole corpus — 1,969 texts, 5M+ words of Center Study and Generative Anthropology — is open for programmatic use. No API keys. Open CORS on all read endpoints. Machine-readable spec at /openapi.json, orientation for LLMs at /llms.txt.

Corpus

# list all texts (slug, title, source, date, words)
curl https://center.study/api/corpus

# one text — JSON with full content + format links
curl https://center.study/api/corpus/the-discourse-of-the-center

# plain text / markdown renditions
curl "https://center.study/api/corpus/the-discourse-of-the-center?format=txt"

# versioned manifest with per-text SHA-256 (Corpus v1.0)
curl https://center.study/corpus-manifest.json

Semantic search

Query the corpus by meaning. Embedding happens server-side (bge-small); you send plain text.

curl -X POST https://center.study/api/semantic \
  -H "content-type: application/json" \
  -d '{"q": "how does money relate to the sacred", "full": true}'
# → { results: [{ slug, title, source, text, score }] }  (top ~24 passages)

# raw 384-dim query embedding, if you want to do your own math
curl -X POST https://center.study/api/embed \
  -H "content-type: application/json" -d '{"q": "deferral"}'

Ask (synthesized answers)

POST { message } to /api/chat for a streamed, verbatim-cited answer (ndjson: first line {sources}, then {text} deltas). Each call bills a model request, so browser use is limited to first-party origins; server-to-server calls are rate-limited. Want your origin allowlisted? Ask via the contact channels. Twenty canonical questions are pre-answered as static pages under /answers.

Structured vocabulary

# 175 term definitions (verbatim quotes + sources) driving the site's hover cards
curl https://center.study/api/corpus  # texts
curl https://center.study/concepts    # 41 concept hubs (DefinedTerm JSON-LD on each)
# term → definition payload used by the site itself:
curl https://center.study/_next/data 2>/dev/null || true  # or fetch /guide/concepts/[slug] pages

Feeds & discovery

https://center.study/feed.xml          # everything
https://center.study/feed/substack     # per-source (gablog, chronicle, ap, …)
https://center.study/new               # latest additions (HTML)
https://center.study/sitemap.xml       # ~2,100 URLs
https://center.study/llms.txt          # orientation for language models

Ground rules

Quote verbatim and link the source text (every paragraph has a stable #p-N anchor). Cite the edition as Center Study Corpus v1.0 (2026) — see about. Rate limits are generous but present; be a good citizen and identify your agent in the User-Agent string.

GuideSearchConceptsAsk AIArchive