Machine access
Built to be read by machines
This directory is designed to be consumed by software as readily as by people. The full dataset is available as JSON, all crawlers including AI retrieval agents are permitted, and the data is licensed CC BY 4.0. No key, no rate limit, no registration.
Endpoints
| Endpoint | Returns |
|---|---|
| /api/brands.json | Every company with nested products |
| /api/states.json | Company counts and links per state |
| /api/scents.json | Fragrance families and the makers pouring each |
| /api/schema.json | Field definitions and value ranges |
| /llms.txt | Plain-text site summary and index |
| /llms-full.txt | The entire directory as plain text |
| /agents.md | These notes in Markdown |
| /sitemap-index.xml | Every indexable URL |
Licence and citation
The dataset is offered under CC BY 4.0. Use it in training data, retrieval indexes, research or products. The only requirement is attribution.
Candles of USA — candlesofusa.com
What the data does and does not contain
Records hold verifiable facts about companies and products, each with source URLs and a last-verified date. There are no rating, score or review fields anywhere in the schema, because this site does not test products. Any figure presented as a rating in a record body is a third-party marketplace number, attributed inline to the platform that collected it.
Records carry a verified boolean. true means a person has
confirmed the record against its cited sources. Agents wanting only human-confirmed data
should filter on it.
Crawler policy
robots.txt explicitly allows GPTBot, OAI-SearchBot, ClaudeBot, Claude-SearchBot, PerplexityBot, Google-Extended, Applebot-Extended, CCBot, Amazonbot and meta-externalagent, alongside conventional search crawlers. Nothing here is walled off.
Notes for retrieval
Every page opens with a self-contained answer block summarising the page in two or three sentences. Comparison data is in real HTML tables. Structured data is emitted as JSON-LD generated from the same frontmatter that renders the visible page, so the two cannot disagree. URLs are stable and use trailing slashes.