Serve at /llms.txt at the root of the domain (e.g. https://docs.example.com/llms.txt).
llms.txt format reference
The llms.txt format, explained.
llms.txt is a small, opinionated Markdown format. Five rules cover almost every valid file in the wild — here they are with examples, plus the parts the spec deliberately leaves open.
Completely free for 30 days. No credit card required.
01Details
Rule 1 — file location
Serve with Content-Type: text/plain or text/markdown. UTF-8.
Same-origin redirects are fine; cross-origin redirects break some agent fetchers and are best avoided.
02Details
Rule 2 — required H1
The first non-blank line must be a Markdown H1: # Project name.
This is the only required element. Everything else is optional but recommended.
03Details
Rule 3 — optional blockquote summary
Immediately after the H1, an optional blockquote (> Short summary in one or two sentences.) gives the agent context before it follows any links.
The spec does not set a hard character limit; in practice we recommend keeping the blockquote under 280 characters because some agent fetchers truncate it.
04Details
Rule 4 — sections of links
Each section starts with an H2: ## Docs / ## API reference / ## Examples.
Inside, a bulleted list of Markdown links: - [Page title](https://example.com/page.md): one-line description.
Always link to the .md version of each page when available. Agents that follow links get clean text instead of HTML chrome.
05Details
Rule 5 — the Optional section
## Optional collects nice-to-have references that the agent can skip when context is tight.
Put marketing pages, changelog entries, deep dives, or anything large here.
Agents that respect the spec will skip Optional first when truncating.
06Details
What the spec leaves open
Section names — pick what fits your domain. There is no enum.
Whether to include rendered code samples inline vs only as links.
Cache headers and freshness — your CDN call.
Whether to ship per-page Markdown files. The spec recommends it; we strongly recommend it for API docs.
07Details
Validation
The official spec lives at llmstxt.org — read it first if you are writing llms.txt from scratch.
Bloom's dashboard previews the parsed structure and flags missing H1, orphaned links, and oversized files.
If you generate from OpenAPI (Bloom, Mintlify, Fern), validation runs every build and stale files do not ship.