# Docs for AI agents

URL: https://www.atmee.ai/docs/v2/getting-started/ai-agents

> Every page of these docs, and the API's contract, as plain files an AI agent can read: an index, the whole docs in one file, a Markdown copy of each page, and the OpenAPI spec.

These docs are written for people and for AI agents. Everything here is also published as plain
files, so a coding agent or a chat assistant can read the docs the way it reads any web page,
without scraping HTML.

Looking for an agent that **makes avatars and builds apps** for you, rather than one that reads
the docs? That is the atmee plugin, its skills and the MCP server:

- [Build with AI agents](/docs/v2/agents): Claude Code, Codex, Cursor or Claude makes the avatar and writes the app around it while you chat.

## The files

| File                   | Address                                             | What it holds                                                                                    |
| ---------------------- | --------------------------------------------------- | ------------------------------------------------------------------------------------------------ |
| **llms.txt**           | `https://www.atmee.ai/docs/v2/llms.txt`             | An index: every page with its address and a one-line summary.                                    |
| **llms-full.txt**      | `https://www.atmee.ai/docs/v2/llms-full.txt`        | Every page in one Markdown file, in sidebar order.                                               |
| **A page as Markdown** | `https://www.atmee.ai/docs/v2/<page>.md`            | One page. Add `.md` to any docs address, such as `https://www.atmee.ai/docs/v2/api/api-keys.md`. |
| **openapi.yaml**       | `https://www.atmee.ai/docs/v2/openapi.yaml`         | The Avatar API's OpenAPI 3.1 spec: every operation an API key can call, with every field.        |
| **Search**             | `https://www.atmee.ai/docs/v2/search?query=<words>` | The docs search: JSON results, each with the page and section it found.                          |

All of these are public and need no sign-in. They change only when the docs do.

## Which to use

* **To answer a question,** start with `llms.txt`, pick the pages that match, and read their `.md`
  copies. This keeps the agent's context small.
* **To load all the docs at once,** read `llms-full.txt`. It is large, around 200 KB, so use it
  when the agent has the room.
* **To write code against the API,** read `openapi.yaml`. It is the source of truth for
  endpoints, fields and error codes: an agent should never call an endpoint or send a field the
  spec does not have. The [API reference](/docs/v2/api/reference/session/createSession) is generated
  from the same file.

## From any docs page

Each page's HTML links its Markdown copy in its head (`<link rel="alternate" type="text/markdown">`).
Under each page's title:

* **Copy Markdown** copies the page as Markdown, ready to paste into a chat or a prompt.
* **Open** opens the page's Markdown, or starts a chat about it in ChatGPT, Claude, Cursor or
  Scira AI.

## Point your agent at the docs

In a coding agent such as Claude Code or Cursor, name the index in your request:

```text
Read https://www.atmee.ai/docs/v2/llms.txt, then use the Atmee docs to add a talking avatar to this
site. Check every API call against https://www.atmee.ai/docs/v2/openapi.yaml.
```

The agent fetches the index, follows it to the pages it needs, and checks its code against the
spec.