Use with AI agents

Feed the whole Behio SDK reference to ChatGPT, Claude, Cursor, Codex and other AI coding tools with a single link.

These docs are built to be read by AI coding agents, not just humans. Every page is available as clean Markdown, and the entire reference is published in machine-readable files that follow the llms.txt convention. Point your AI tool at one link and it can pull in everything the SDK supports.

The links to give your AI

Link	What it is
`https://sdk.behio.com/llms.txt`	A compact index: every page with a one-line description and its URL. Best when the tool fetches pages on demand.
`https://sdk.behio.com/llms-full.txt`	The complete documentation in one Markdown file (every guide, every signature). Best for "ingest everything at once".
`https://sdk.behio.com/docs/<page>` + `content.md`	Any single page as raw Markdown. Example: `https://sdk.behio.com/llms.mdx/docs/catalog/products/content.md`

The files are generated from the same source as this site, so they update automatically when the docs change. Nothing to maintain by hand.

How to use it

ChatGPT / Claude

Start a project (Claude Projects or a ChatGPT project), then add https://sdk.behio.com/llms-full.txt as a source, or paste the URL into the chat and ask it to read it. The model then has the full SDK reference in context and can scaffold a storefront, wire hooks, or debug calls against the real API.

Cursor / Windsurf

Add the docs as a custom documentation source (Cursor: Settings to "Docs" to "Add new doc") using https://sdk.behio.com/llms-full.txt or the site URL. Then reference it in chat with @Behio so completions are grounded in the SDK.

Codex and other coding agents

Give the agent https://sdk.behio.com/llms-full.txt as a reference at the start of the task ("Read this first, then build the storefront"). One link is enough, the file embeds every example and type, so the agent does not have to crawl.

Per-page Markdown

On any docs page, use the Copy Markdown button (top of the page) or the View options menu to grab that single page as Markdown, or open its content.md URL directly. Handy when you only need one topic, like Currency or Checkout.

MCP server (live connection)

For agents that speak the Model Context Protocol (Claude, Cursor, Cline, and others), there is a hosted MCP server so the agent can search and read the docs as live tools, instead of ingesting a static file:

https://sdk.behio.com/mcp

It exposes three tools:

Tool	What it does
`list_docs`	List every documentation page with its title, URL and description.
`search_docs`	Search the docs for a topic and get back the most relevant pages.
`get_doc`	Fetch the full Markdown of a single page by its path.

Connect it

Claude Code:

claude mcp add --transport http behio https://sdk.behio.com/mcp

Cursor / Claude Desktop / other clients, add to the MCP servers config:

{
  "mcpServers": {
    "behio": {
      "url": "https://sdk.behio.com/mcp"
    }
  }
}

The server is read-only and needs no authentication.

This is the docs server, not your account

sdk.behio.com/mcp only reads this documentation. To let an assistant work with your real data (warehouse, products, orders, invoices, contacts), connect the Behio account server at be.behio.com/mcp instead. See Connect your AI assistant.

What the AI gets

Because it reads the same docs you do, the agent knows the full surface: catalog and search, cart and checkout, customer and orders, the React hooks, currency switching, per-domain product scoping, and every type. Treat the link as the canonical reference and you get accurate code with far fewer hallucinated methods.

Use with AI agents

On this page