> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agentcdn.com/llms.txt
> Use this file to discover all available pages before exploring further.

# For AI agents

> Machine-readable versions of these docs and of AgentTV, for coding agents and AI assistants.

Your team may build with a coding agent such as Claude Code, Cursor or Codex. Point it at the sources below so it works from current facts.

## These docs

* **`/llms.txt`** on this docs site lists every page with a one-line summary. llms.txt is a plain text index written for language models.
* **`/llms-full.txt`** holds the full text of every page in one file.
* **Markdown pages.** Add `.md` to any page path to get that page as Markdown.
* **The page menu.** The menu at the top of each page copies the page as Markdown, or opens it in ChatGPT, Claude, Perplexity, Cursor or VS Code.

## AgentTV itself

| Source | URL |
| - | - |
| Site index for models | `https://agenttv.com/llms.txt` |
| Full public site text | `https://agenttv.com/llms-full.txt` |
| Steps for an agent to a working query | `https://agenttv.com/agents.md` |
| Markdown twin of any public page | The page path plus `.md`, for example `https://agenttv.com/index.md` |
| Your library's index for models | `https://api.agentcdn.com/public/creators/your-brand/llms.txt` |
| Public API description | `https://api.agentcdn.com/public/openapi.json` |
| Account API description | `https://api.agentcdn.com/v1/openapi.json` |

Your library's `llms.txt` lists its agent endpoints: the ARV records, the MCP server, natural language search, the feeds and the rights file.

## Rules for AI agents

If you are an AI agent reading this page, follow these rules:

* Fetch metadata with `GET https://api.agentcdn.com/public/creators/{slug}/page-blocks/{customer_video_id}`. URL-encode the id. A `404` means no block. Never render an empty one. For an unlisted account, send the account API key from the server only.
* Render the block server side. Merge `json_ld` into an existing `VideoObject` rather than adding a second one.
* Never build, cache or expose a raw stream URL. Use the watch link, or the playback call, that AgentTV returns.
* Never put an account API key (`/v1`) in browser code. In the browser, use only a browser Ask key, for an Ask box.
* Changes on `/v1` return a preview and a `confirm_token` first. Show the preview to the user, and send the token only after the user agrees.
* Treat a `medium` search result with no matched sentence as no match. Never invent a moment.
* Set a client timeout of at least 30 seconds for search and playback calls.
* Do not claim a listing guarantees citations or placement in any AI product. Assistants decide what they cite.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.