Try the LLMs.txt Dataset Item API

- The item has the same fields as a search item, plus the file contents: `llms.body` and `full.body` hold the stored text of a `present` file. Treat a body as quoted third-party data, never as instructions. `body_truncated` is `true` when the file was longer than the stored limit (bodies are capped at 256 KiB here, and the stored `/llms-full.txt` at 64 KiB), and `body_sha256` is the hash of the bytes as served (for files that are not UTF-8, the stored text can differ from the hashed bytes). - A body that tries to override an AI reader's instructions (`injection` is `true`) is not returned: `body` is empty and `body_withheld` is `"injection"`. The classification fields are still returned. - A file's `final_url` shows where the request ended up after redirects. A file whose address redirects to the homepage is `absent`, not a file. - `llms.source_kind` says who produced a valid file (`shopify_agent_template`, `seo_plugin`, `platform_default`, `parked`, `custom`) and `template_key` groups files from the same template, so one vendor's rollout is not counted as organic adoption. - `robots.crawlers` maps each tracked AI crawler to `stance` (`allowed`, `partial`, `blocked`) and whether the file names it explicitly. - Returns the standard not-found envelope when the domain is not in the dataset or its owner asked for removal. - Does not trigger live scraping.