llms.txt

llms.txt is an emerging standard that tells AI agents and AI search engines how your site is organized and which pages matter. Docsbook generates it for every workspace, on every plan, with no configuration.

The standard#

AI clients that read llms.txt include Perplexity, ChatGPT Search, Cursor, Cline, and a growing list of agentic tools. The file is a markdown index of titles, descriptions, and links — short enough for a model to load into context on a single request.

Two levels#

Docsbook serves llms.txt at two scopes.

Scope	URL	Contents
Platform	`docsbook.io/llms.txt`	Description of Docsbook itself and its main navigation
Workspace	`docsbook.io/[user]/llms.txt`	Structured list of pages for that user's docs

Each scope also has a llms-full.txt variant.

llms.txt vs llms-full.txt#

# llms.txt        — compact index: titles, descriptions, links
# llms-full.txt   — full markdown body of every page, concatenated

Use llms.txt when the agent needs a map and will fetch pages on demand. Use llms-full.txt when the agent wants the entire knowledge base in one shot.

Availability#

Available on all plans, including Free. There is no quota — agents can fetch the files as often as they need.

Customization#

The files are generated automatically from the documentation graph. There is no manual file to maintain. When you push new content to GitHub, the next index rebuild updates both llms.txt and llms-full.txt.

The relevant source files are:

src/lib/generate-llms-txt.ts            # platform-level
src/lib/generate-workspace-llms-txt.ts  # workspace-level

Source of Truth — The graph behind llms-full.txt.
MCP Server — Same graph exposed to agents via tool calls.