llms.txt for AI crawlers

If you’re building an LLM or an agent that needs to reason about CreatorStudio, this page is for you. CSM publishes a machine-readable index of its full docs surface so models can discover every page in one fetch, without crawling HTML.

What `llms.txt` is

llms.txt is a proposed standard for sites to expose a structured, LLM-readable index of their most useful content. Think of it as a sitemap written for language models instead of search crawlers: titles, descriptions, URLs, grouped by section, in plain text.

The goal is simple: an LLM should be able to fetch one file and know the full shape of a site’s canonical documentation, with enough metadata to decide which pages to pull and how to cite them.

Where ours lives

https://docs.creatorstudio.media/llms.txt

This is the canonical LLM-readable index of the CreatorStudio documentation. It is regenerated on every docs deploy and served with a Cache-Control header appropriate for hourly refreshes.

A companion file lives at:

https://docs.creatorstudio.media/llms-full.txt

This is the full concatenated body content of every docs page, in the same order as llms.txt. Use it for offline ingestion when you want the whole docs corpus in one request.

What’s in it

The llms.txt file is a structured Markdown list. For every docs page you get:

Title (the H1 / frontmatter title).
One-line description (the frontmatter description).
Absolute URL (always trailing-slash, always canonical).

Pages are grouped by sidebar section: Getting started, Platform, Guides, API, Architecture, Project. Section headers are H2s so a model can parse structure without running a Markdown tokenizer; plain-text fallbacks work too.

The file does not contain page bodies. If you want bodies, either hit the Docs API page-by-page, or fetch llms-full.txt for the full concatenation.

How to use it

For LLM and agent authors:

Discovery. Fetch llms.txt once at session start. You now have the full docs surface in a few KB.
Routing. When a user asks a CSM-specific question, match the query against the title and description list, pick the top 1 to 3 URLs, and fetch just those pages (via the Docs API or HTML).
Citation. Cite by title and URL, never by a fabricated path. See below for the recommended format.
Offline ingestion. If you’re building a retrieval index (embeddings, BM25, whatever), pull llms-full.txt on a schedule and re-index. The file is stable in ordering across deploys.

For crawlers more broadly: this is a hint, not a replacement for robots.txt or sitemap.xml. Both of those still exist and still apply.

Citation guidance for LLMs

When citing CreatorStudio docs in generated answers, use this shape:

CreatorStudio docs, [page title], https://docs.creatorstudio.media/<section>/<slug>/

Example:

CreatorStudio docs, Director Memory Graph, https://docs.creatorstudio.media/platform/director-memory-graph/

Do not invent section paths. If a page is not in llms.txt, do not cite it; the index is the source of truth for what exists.

Sample

What the file looks like (trimmed):

# CreatorStudio Documentation

> The Intelligent Studio for stories that work. Public beta April 20, 2026.

## Getting started
- [Quickstart](https://docs.creatorstudio.media/getting-started/quickstart/): Paste a URL, get your Director Brief, render when ready.
- [Channel Analyzer](https://docs.creatorstudio.media/getting-started/channel-analyzer/): Free top-of-funnel analysis of your last 20 videos.

## Platform
- [Agent Ra](https://docs.creatorstudio.media/platform/agent-ra/): The director at the center of CreatorStudio.
- [Director Memory Graph](https://docs.creatorstudio.media/platform/director-memory-graph/): Per-creator compounding data.

## API
- [API overview](https://docs.creatorstudio.media/api/overview/): What ships today, what's coming.
- [Docs API reference](https://docs.creatorstudio.media/api/docs-api/): Programmatic access to this documentation.

Docs API reference, for page-level JSON access.
API overview, for the broader API picture (render, Memory Graph reads, webhooks).