Connect Stell's docs to ChatGPT, Claude or your coding agent

If you build against an API in 2026, there is a good chance your first question goes to an AI assistant, not a search bar. That works well right up until the assistant answers from memory: training data that is months old, blended with documentation from other products, delivered with complete confidence. You paste the suggested payload, the API returns a validation error, and you spend twenty minutes discovering the assistant invented a field.

There is a better way to wire this up. Stell's documentation at docs.getstell.com is available as an MCP server. Connect it to your AI of choice and every answer about Stell comes from the live docs and API reference, current as of the moment you ask.

What MCP is, briefly

MCP, the Model Context Protocol, is an open standard that lets AI assistants call external tools and read external data sources. Think of it as a USB port for AI: any assistant that speaks the protocol can plug into any server that offers it.

A docs MCP server gives the assistant one focused capability: search the real documentation. When you ask "how do I update a pass through the Stell API", the assistant queries the docs, reads the actual endpoint reference, and answers from that, with links back to the source. No guessing, no stale payloads, no fields that were renamed two releases ago.

This matters more for Stell than for most APIs, because the surface keeps moving in ways training data cannot follow. Wallet platforms ship new capabilities every year, and Stell's job is to expose them quickly: new pass layouts, new fields, new webhook events. The documentation tracks all of it. An assistant reading the live docs inherits that currency for free.

The endpoint

Everything below uses one URL:

https://docs.getstell.com/mcp

It is a streamable HTTP server with no authentication, so connecting takes a minute in any MCP-capable client.

Connect it to Claude

In Claude, open settings, go to connectors, and choose to add a custom connector. Give it a name like "Stell docs" and paste the endpoint URL. Once added, Claude can search Stell's documentation in any conversation, on web, desktop, and mobile.

Connect it to ChatGPT

In ChatGPT, open settings, go to connectors, and enable developer mode under advanced settings if you have not already. Then create a new connector, name it, and paste the endpoint URL with no authentication. ChatGPT will pick up the docs search tool and use it when you ask about Stell.

Connect it to your coding agent

This is where it earns its keep. An agent writing your integration can check the reference for itself while it works.

For Claude Code, one command:

claude mcp add --transport http stell-docs https://docs.getstell.com/mcp

For Cursor, VS Code, Windsurf, and other MCP-capable editors, add it to your MCP configuration:

{
  "mcpServers": {
    "stell-docs": {
      "url": "https://docs.getstell.com/mcp"
    }
  }
}

From then on, when you ask the agent to build a webhook handler or wire up pass issuance, it reads the current API reference before writing a line. The difference shows up in the details agents usually get wrong: exact field names, required headers, the shape of a webhook delivery. An agent grounded in the live reference writes the integration the API actually accepts, which shortens the part of a build-or-buy decision that is really about how long integration takes.

Why this beats copy-pasting docs

You could paste documentation pages into a chat, and people do. The MCP route is better on three counts.

It is always current. The server searches the docs as they exist right now. When we ship a new endpoint or field, your assistant knows about it the same day, with no action from you.

It is complete. The assistant searches the whole documentation set, guides and API reference alike, rather than the two pages you thought to paste. Questions that span topics, like how pass lifecycle events connect to webhook deliveries, get answers drawn from both.

It is set once. The connector persists across conversations. Every future question about Stell, from anyone on your team who sets it up, gets grounded answers.

What this says about the platform

A docs MCP server is a small thing, but it reflects how Stell thinks about integration: the fastest path from "we should do wallet passes" to passes in wallets. The same instinct is behind the portal for teams who never want to touch the API, and the API and webhooks for teams who want to automate everything. However you build, the documentation should meet you where you work, and in 2026 that includes inside your AI assistant.

The short version

• Stell's documentation is available as an MCP server at https://docs.getstell.com/mcp, no authentication required.
• Connect it to Claude or ChatGPT through their connector settings, or to coding agents like Claude Code and Cursor through their MCP configuration.
• Once connected, your assistant answers Stell questions from the live docs and API reference instead of stale training data.
• It takes a minute to set up and applies to every conversation afterwards.

Connect it before your next integration question. The answer will come from the docs, not from memory.