LLM-optimized documentation
Every page on this site is also available as clean Markdown, and the whole site is indexed
for AI coding agents such as
Claude Code,
Cursor, and
Windsurf.
The files follow the
llms.txt convention, so AI search engines find
them too.
There are three levels of granularity:
| What you get | Where | Use it when |
|---|---|---|
| One page as Markdown | append .md to the page URL |
an agent needs one page |
| An index of every page | /docs/v2/union/llms.txt |
an agent needs to find the right page first |
| The entire site in one file | /docs/v2/union/llms-full.txt |
you need everything at once |
Open a page in Claude or ChatGPT
Every page has a menu at the top right, on the breadcrumb row. Open in Claude hands the page’s Markdown URL to Claude with a prompt asking it to read the page so you can ask questions about it. The dropdown beside it offers:
- Open in Claude and Open in ChatGPT: the same action, for either assistant.
- Copy page: copies the page’s Markdown text to your clipboard, for pasting into a tool that cannot fetch URLs.
- View as Markdown: opens the page’s Markdown version in the browser.
One page as Markdown
Append .md to any page URL. For example, this page is at:
https://www.union.ai/docs/v2/union/api-reference/flyte-context/
and its Markdown version is at:
The same rule applies to section landing pages: .../user-guide/tasks/ becomes
.../user-guide/tasks.md. A section landing page’s Markdown ends with a Subpages list of
every page directly beneath it, with each page’s URL and headings, so one fetch tells an agent
what the section contains and which page to read next.
Two details worth knowing:
- Each Markdown file starts with a short identity block naming the product, the version line, and the index URL, so a model knows what it is reading even when it is handed the file with no other context.
- The variant root has no Markdown version of its own. Appending
.mdto the root URL redirects to that variant’sllms.txt, which is the index for the whole tree: `/docs/v2/union.md` redirects to `/docs/v2/union/llms.txt`.
You can also request Markdown at the page’s own URL by sending an Accept: text/markdown
header:
$ curl -H "Accept: text/markdown" https://www.union.ai/docs/v2/union/user-guide/tasks/All links inside the Markdown files are absolute (https://www.union.ai/docs/...), so a file
keeps working after it is copied out of the site.
An index of every page (llms.txt)
llms.txt lists every page on the site, grouped by section, with the H2 and H3 headings found
on each page. An agent can pick the right page from the index and then fetch only that page’s
Markdown.
-
/docs/v2/union/llms.txt(about 50K tokens)
To give a coding agent standing access to the docs, append the index to the AGENTS.md,
CLAUDE.md, or equivalent file in your project root. Put it in a file the tool loads into
context by default. Adding it as a skill or tool works less well, because the agent then has
to decide to load it rather than always having it available.
$ curl https://www.union.ai/docs/v2/union/llms.txt >> AGENTS.mdYou are viewing the Union.ai docs.
To get the llms.txt for a different product variant, use the variant selector at the top of the page.
The entire site in one file (llms-full.txt)
llms-full.txt is the whole documentation set in a single file. It is several megabytes, so
prefer llms.txt plus the pages it points to unless you need everything at once.