A practical check for file identity, structure, redirects, links, and the limits of what one retrieval can establish.
A server response can be successful while the body is an HTML page, an unrelated file or a misleading response. Confirm the retrieved document itself before you describe the endpoint as an llms.txt example.
Steps to use
Fetch the proposed endpoint and save the response status, content type, requested URL and final URL.
Read the start of the returned body. Check for a Markdown title and actual Markdown links; do not rely on the filename or a third-party directory listing.
Resolve relative links against the final URL and sample important destinations separately. A link count alone does not prove those destinations work.
Date the check. Keep structure measurements separate from claims about quality, model behavior, search rankings or AI citations.
Where judgment matters
This directory checked the llms.txt files directly, but it did not recursively validate their linked documentation or test how agents use them. Availability is a narrow observation, not an effectiveness result.
A developer documentation map retrieved at the docs hostname and redirected to the current platform.claude.com endpoint. It includes language and topic routes into the API docs.
The retrieved Markdown is mid-sized (81,097 characters), with 3 second-level headings and 749 Markdown links. These are descriptive counts, not a quality score.
A deeply grouped documentation map with separate paths for concepts, quickstarts, functions and data features.
The retrieved Markdown is mid-sized (46,358 characters), with 33 second-level headings and 406 Markdown links. These are descriptive counts, not a quality score.
A version-labeled search documentation map organized into overview, API reference, guides and cloud management.
The retrieved Markdown is mid-sized (30,788 characters), with 6 second-level headings and 135 Markdown links. These are descriptive counts, not a quality score.