Search
The agent asks for the documentation relevant to its current task.
Documentation for AI agents
Give every reader a direct path to a citable answer.
SaturnDocs publishes one documentation source through a human-readable site, llms.txt, an MCP server, search, and a cited AI assistant. Human readers and AI agents reach the same current material.
The direct answer
What is developer documentation for AI agents?
Developer documentation for AI agents is structured technical content with direct machine-readable discovery and retrieval paths. SaturnDocs provides those paths through llms.txt, an MCP server, stable page URLs, clear heading structure, and an assistant that answers from published pages and cites the sources behind each answer.
The goal is one durable body of documentation rather than a separate set of answers for machines. A correction to the source improves the website, search, the assistant, llms.txt, and MCP retrieval together.
Discovery
How do AI agents find the documentation?
Every SaturnDocs customer site publishes llms.txt as a concise index and llms-full.txt as consolidated context. These files identify the organization, describe the documentation, and link to canonical pages. Visible links make the files discoverable to agents that browse the site instead of probing known root paths.
The files are generated from the same content data that drives the site, so feature descriptions, page links, and canonical URLs do not drift into a second hand-maintained version.
Retrieval
How do AI agents search the documentation while they work?
The documentation MCP server gives compatible agents a structured search path. An agent can ask for a setup step, endpoint, configuration field, or error, retrieve the relevant published material, and keep the source location attached to the result.
Retrieve
The server returns material from the current published source.
Cite
The answer keeps the page location available for verification and deeper context.
Answer quality
Why should documentation answers include citations?
A citation makes an answer inspectable. It shows which published page supports the response, lets the reader inspect prerequisites and warnings, and exposes missing documentation when no source contains the answer. SaturnDocs treats that missing answer as a specific page or revision to produce.
Read the detailed explanation in Inside the docs assistant, or see every included machine surface on What you get.
AI-agent documentation questions
Frequently asked questions.
What is llms.txt?
llms.txt is a plain-text map of a documentation site for AI tools. It identifies the organization and links to durable pages with short descriptions. SaturnDocs publishes it automatically from the same source as the human-readable site.
What does the documentation MCP server do?
The MCP server gives compatible AI agents a structured way to search the documentation and retrieve relevant material while they work. It returns information from the current documentation source rather than from a separate hand-maintained copy.
How does the SaturnDocs assistant avoid guessing?
The assistant answers from the published documentation and cites the pages that support its response. If the documentation does not contain enough information, the assistant says so instead of filling the gap with an unsupported answer.
Can AI agents access private documentation?
Only when they use an authenticated path that grants access to the protected material. Public machine-readable surfaces expose public documentation; private content remains subject to the same reader permissions as the site.
One source, every reader
Make your product understandable to people and their agents.
Get your docs done