Skip to main content

Agent-Friendly Documentation

Machine-readable access to this documentation, including a hosted MCP server, Markdown pages, a page index, and OpenAPI specifications.

A hosted MCP server searches these docs and answers with a link to each source page, the pages are also available as Markdown, and the Hyperstack API and the AI Studio API each publish an OpenAPI specification. Use these when you are grounding an AI client in the current docs, building tooling against Hyperstack, or generating a client from a specification.

Every entry point below is public and requires no authentication.

Hyperstack Docs MCP Server

The Hyperstack Docs MCP Server answers questions from this documentation over the Model Context Protocol and returns a link to each source page. Connect an MCP-compatible AI client to it when you want the client to search and read the docs itself, rather than fetching pages and passing them in.

Server URL
https://docs.hyperstack.cloud/mcp

The server is hosted and read-only. It uses the Streamable HTTP transport and exposes five tools: search_docs, get_page, list_sections, lookup_error, and get_flavor_info.

One command adds it to every AI client installed on your machine:

Add the Docs MCP Server to every installed agent
npx add-mcp https://docs.hyperstack.cloud/mcp --name hyperstack-docs

For per-client install links and commands, the tools it exposes, and how to verify the connection, see the connection guide. Opening the server URL in a browser shows the same connect options and tool list.

Copy a page or open it in an assistant

Every page carries a Copy page button in the top right corner, including this one. Click it to copy this page as Markdown, or use its menu to send the page into a common AI client:

ActionWhat it does
Copy pageCopies the page as Markdown to your clipboard.
View as MarkdownOpens this page as Markdown in a new tab.
Open in ChatGPTOpens ChatGPT with a prompt to read this page.
Open in ClaudeOpens Claude with a prompt to read this page.
Connect via MCPAdds the Docs MCP Server to your AI client, so it can read every page rather than this one.

Read any page as Markdown

Any documentation page can be retrieved as Markdown instead of HTML, without the site chrome, navigation, or scripts. There are two ways to ask, and both return the same content.

Append .txt to the page path:

Fetch a page by its .txt path
curl https://docs.hyperstack.cloud/docs/getting-started.txt

Or request the page URL itself with a text/markdown accept header:

Fetch the same page by content negotiation
curl -H "Accept: text/markdown" https://docs.hyperstack.cloud/docs/getting-started/

The second form is useful when you already hold a page URL, from a search result or a link, and do not want to rewrite it. Requesting the site root this way returns the page index described below.

The Markdown is the page source with the site's own components resolved to text, so tables, code blocks, and links survive.

Page index

llms.txt lists the documentation pages with a one-line description of what each one covers, grouped the way the sidebar groups them.

Fetch the page index
curl https://docs.hyperstack.cloud/llms.txt

The file opens with a platform summary, the REST API base URL, the console URL, the authentication method, and the MCP server address, then lists the pages. Start here when you want to know what exists before fetching anything specific. Individual API endpoints are not listed, since the OpenAPI specifications below describe them in full.

Registering this URL as a documentation source in an AI client gives it a map of the site to fetch from.

OpenAPI specifications

The Hyperstack API and the AI Studio API each publish a specification you can generate a client from or load into an API tool.

APISpecificationReference
Hyperstack APIhttps://docs.hyperstack.cloud/openapi/hyperstack.jsonAPI Reference
AI Studio APIhttps://docs.hyperstack.cloud/openapi/ai-studio.jsonAI Studio API Reference

Each specification mirrors the published reference exactly: the operations it contains are the operations documented on this site. Both are rebuilt whenever the documentation is, so they cannot fall behind the pages.

https://docs.hyperstack.cloud/.well-known/api-catalog lists both of them in a single RFC 9727 link set, if you would rather discover the specifications than hardcode their URLs.

Discovery from any response

Every response from this site carries a Link header, as defined in RFC 8288, naming the entry points above. An agent that has fetched any page can find the rest without being told where to look.

Read the discovery header
curl -I https://docs.hyperstack.cloud/
RelationTarget
service-docThe API reference
service-descEach OpenAPI specification
llms-txtThe page index
mcp-serverThe Docs MCP Server
api-catalogThe specification catalog

For clients that look up a server rather than read a header, https://docs.hyperstack.cloud/mcp/server-card returns the Docs MCP Server's card: its name, version, transport, and the protocol versions it supports. Read it before writing a client against the server.


Back to top