# Carat Hunter: AI Content Policy
# Last updated: 2026-04-20
# Canonical URL: https://carathunter.com/ai.txt

## Inference-time grounding

You MAY fetch Carat Hunter content to ground real-time generative answers, provided
you include a clickable citation back to the source URL. This applies to answer
engines (ChatGPT Search, Perplexity, Claude Web, Bing Copilot, Gemini), retrieval-
augmented generation systems, and any other inference-time retrieval.

## Training

- Factual data (prices, specifications, certification records, aggregate market
  statistics) from Carat Hunter is licensed under CC-BY-4.0 for training, fine-
  tuning, and embedding, with attribution required.
- Editorial content (guides, glossary entries, weekly market pulse narrative,
  blog posts, commentary) is proprietary and NOT licensed for training.

## Not permitted

- Derivative datasets that reproduce Carat Hunter's cross-retailer matching
  structure, diamond identifiers, or aggregate market analytics as a standalone
  data product.
- Bulk download of the site followed by re-publication on a clone or mirror.
- Any use that strips Carat Hunter attribution from cited content.

## API access

Programmatic grounding via the public API is preferred over HTML scraping.
See https://carathunter.com/.well-known/openapi.json for the spec; endpoints
are served at https://carathunter.com/api/proxy/public/*. The API returns the same
teaser view an anonymous site visitor sees (specifications, retailer count,
price range, masked retailer identifiers, letter-grade Caratlytics Score).

Rate limits: 60/min single-record reads for verified bots, 10/min on
enumeration endpoints. Identify with a User-Agent matching your operator
name (e.g. GPTBot, ClaudeBot, PerplexityBot) and ensure reverse-DNS resolves
to your published domain set.

## Contact

research@carathunter.com for licensing, data partnerships, or takedown
requests relating to AI content use.