llms.txt release notes

v2 — 2026-08-10

The proposal’s Changes page explains the revision as a response to adoption: “Thousands of sites now publish an llms.txt file, documentation platforms generate one automatically, and coding agents use them reliably. That shift is what this revision reflects.” The file format, from the optional BOM to the H2 file lists, is unchanged.

• Discoverability through link relations. rel="alternate" type="text/markdown" points to a page’s markdown version and rel="describedby" to the llms.txt that covers it, as HTML <link> elements or an HTTP Link: header. See Markdown page versions and link relations.
• Both markdown URL forms. v1 specified .md appended to the full URL (page.html.md). v2 also allows the extension replaced (page.md), and index.md alongside index.html.md for URLs without a file name.
• Subpath files defined. v1 permitted files in subpaths “without saying what that meant”. v2: a file covers the pages under its path, and the most specific file applies.
• Consumption model stated. v1 said nothing about how the file should be used. v2: agents view or search the llms.txt, then follow the relevant links, which should point to LLM-friendly content.
• `llms_txt2ctx` and the mechanical meaning of Optional removed. The context-expansion tooling is no longer part of the proposal. ## Optional sections are still allowed as a convention for secondary links.
• Background and examples rewritten to describe how agents use websites now, rather than predicting that they might.

The context-expansion tooling is no longer part of the proposal, and with it goes the special meaning of the `Optional` section, which told those tools what to omit. Optional sections are still allowed, and remain a useful convention for secondary links, but they no longer carry mechanical semantics.

v1 permitted llms.txt files in subpaths without saying what that meant. v2 defines it: a file covers the pages under its path, and the most specific file applies. This is also what lets a site that only controls a path, such as a GitHub Pages project site, participate fully.

Comparing the v1 and v2 texts shows a few further changes that the Changes page doesn’t list separately:

• The last guideline now reads “Test your file by asking an agent questions about your content, giving it only your llms.txt as a starting point.” v1’s version asked authors to run a tool that expands the file into an LLM context and test language models against it.
• A paragraph explains why llms.txt doesn’t use RFC 8615’s /.well-known/ prefix: well-known URIs exist only at the origin root, and many authors control only a path.
• The background names adopters: llms.txt files from OpenAI, Anthropic and Gemini for their developer docs, and Chrome’s Lighthouse, which audits sites for one.
• The integrations list adds Mintlify, GitBook, Yoast SEO, AIOSEO and Wix, and an MCP server for fetching and searching llms.txt files. llms_txt2ctx is no longer listed. llmstxthub.com joins the directories.

Optional byte-order mark — 2026-06-09

A change to the v1 text, recorded in the proposal’s repository: “An optional byte-order mark (BOM)” was added as the first item in the ordered list of elements. A file may start with a BOM; it is never required. v2 keeps the line.

v1 — 2024-09-03

The original proposal, by Jeremy Howard: “A proposal to standardise on using an /llms.txt file to provide information to help LLMs use a website at inference time.” Quotes below are from the v1 text as it stood just before v2, which is the version in the repository’s history.

• Location: “files located in the root path /llms.txt of a website (or, optionally, in a subpath)”. What a subpath file covered was not defined.
• Format: the same ordered elements as today: an H1 (“the only required section”), a summary blockquote, body text, and H2 “file lists” of [name](url) links with optional notes.
• Markdown versions: pages provide “a clean markdown version of those pages at the same URL as the original page, but with .md appended. (URLs without file names should append index.html.md instead.)”
• Processing: “This proposal does not include any particular recommendation for how to process the llms.txt file, since it will depend on the application.” It described FastHTML expanding its file into llms-ctx.txt and llms-ctx-full.txt with the llms_txt2ctx command line tool.
• Inference, not training: “Our expectation is that llms.txt will mainly be useful for *inference*, i.e. at the time a user is seeking assistance, as opposed to for *training*.”

Note that the “Optional” section has a special meaning—if it’s included, the URLs provided there can be skipped if a shorter context is needed. Use it for secondary information which can often be skipped.

Neither v1 nor v2 defines llms-full.txt. That filename is a convention some platforms adopted, not part of the proposal.

LLMSRank rubric releases

Since v2.0 the rubric follows llms.txt v2, and every score also records the llms.txt version it was judged against. The full history, with the effect of each change on our corpus, is in the rubric changelog.

Sources

• llmstxt.org — Changes: v2 (August 2026)
• llmstxt.org — The /llms.txt file, v2 (modified 2026-08-10)
• llmstxt.org — The /llms.txt file, v2, markdown version
• AnswerDotAI/llms-txt — nbs/index.qmd, the v1 text as of 2026-06-09
• AnswerDotAI/llms-txt — the proposal’s GitHub repository

Related

• About llms.txt
• Markdown page versions and link relations
• Writing an llms.txt file