Editorial policy
Authority depends on a reader being able to tell fact, observation, proposal and interpretation apart. These are the rules used across llmtxt.info.
Last updated:
Source hierarchy
- Primary sources: an official specification, vendor documentation, release note, source repository or served public response.
- Reproducible observation: a dated llmtxt.info measurement with a published method and downloadable result.
- Secondary reporting: used for context, never as the only proof of a technical behavior when a primary source is available.
Claims with no inspectable source are removed, qualified, or marked as an opinion.
Evidence labels
- Provider documented: the publisher states the behavior in its own documentation.
- Observed by llmtxt.info: a dated public response or measurement supports a narrow observation.
- User-triggered workflow: a documented manual action can use a file. This does not imply automatic discovery.
- Specification proposal: a convention describes an option. It does not demonstrate client support.
- Unknown or unverified: the site does not infer an answer from a logo, a public file, or marketing copy.
Independence and disclosures
llmtxt.info is not affiliated with llmstxt.org, Answer.AI, or an AI provider. A company
appearing in a directory, example, dataset or article is a source reference, not a customer,
partner or endorsement. The site does not sell placements. If this changes, the placement will
be labeled and the relevant link will use rel="sponsored".
No logo is used to imply a relationship that is not documented. External citations are listed only when the original third-party page can be linked directly.
Corrections
Readers can report a factual error, stale file or missing qualification at [email protected]. Material corrections are recorded in the public corrections ledger with the affected page, date and nature of the change. The changelog records changes to tools, data and method separately.