---
title: Agent setup
description: How to point a coding agent at this documentation.
---

Every page in this documentation, other than the front page, has a
plain-Markdown copy: drop the trailing slash from the page's URL and add
`.md`. For example, this page is at `/docs/agent-setup/`, and its Markdown
copy is at `/docs/agent-setup.md`. Requesting that URL returns the page's
title, description and body as Markdown, with no page chrome (navigation,
header or footer) around it. The same rule applies to blog posts: drop the
trailing slash and add `.md`, for example `/blog/posts/<post>.md`.

`/docs/llms.txt` lists the `.md` URL of every page in the navigation,
grouped the same way the navigation is. `/blog/llms.txt` lists every
published blog post's page.

To hand this documentation to a coding agent, do one of the following:

- Give the agent a single page's `.md` URL directly, when the task only
  concerns that page.
- Give the agent the `/docs/llms.txt` URL, and let it fetch whichever `.md`
  URLs from that list it needs.
- Give the agent both `/docs/llms.txt` and `/blog/llms.txt` when the task
  also touches blog content.

An agent that can fetch URLs can use either approach directly. An agent that
cannot fetch URLs still benefits from being given the raw Markdown pasted
into its context, since the `.md` copy has no navigation or layout markup to
strip out first.
