---
title: Build with AI
description: Point Claude, ChatGPT, or Cursor at the RizPay docs MCP and build against the API faster
---


RizPay's documentation is AI-native. You can point your AI coding tools straight at these docs and have them read, search, and reason over the RizPay API while you build - no copy-pasting pages into a chat window.

The fastest way is the RizPay Docs MCP server. It lets tools like Claude, ChatGPT, and Cursor search these docs and pull full pages on demand, so your assistant answers with real RizPay reference material instead of guessing.

## The RizPay Docs MCP

[MCP](https://modelcontextprotocol.io) (the Model Context Protocol) is an open standard that lets AI tools connect to external data sources. RizPay runs a small, read-only MCP server over these docs.

|               |                                                        |
| ------------- | ------------------------------------------------------ |
| **Endpoint**  | `https://www.rizpay.app/api/mcp`                       |
| **Transport** | Streamable HTTP                                        |
| **Auth**      | None - the docs are public                             |
| **Scope**     | Read-only. It only searches and returns documentation. |

It exposes two tools:

- **`search_docs(query, source?, limit?)`** - Search the developer and storefront docs. Returns ranked results, each with a title, source, URL, markdown URL, and snippet.
  - `query` (required) - your search terms.
  - `source` (optional) - limit to one docs source: `developers` or `storefront`.
  - `limit` (optional) - max results, default 8, max 25.
- **`get_doc(source, slug)`** - Fetch the full markdown of one page.
  - `source` (required) - `developers` or `storefront`.
  - `slug` (required) - the path within that source, e.g. `getting-started/quickstart` or `reference/error-codes`.

A typical flow: your assistant calls `search_docs` to find the right page, then `get_doc` to read it in full before writing code against the API.

## Set it up in your AI tool

Pick your client below. The endpoint is the same everywhere: `https://www.rizpay.app/api/mcp`. There is no API key to configure - the docs are public.

### Claude Code

Add the server with a single command:

```bash
claude mcp add --transport http rizpay-docs https://www.rizpay.app/api/mcp
```

Then check it registered:

```bash
claude mcp list
```

Prefer to configure it by hand? Add this to your `.mcp.json` (project scope) or your user config:

```json
{
  "mcpServers": {
    "rizpay-docs": {
      "type": "http",
      "url": "https://www.rizpay.app/api/mcp"
    }
  }
}
```

### Claude Desktop and claude.ai

Add RizPay as a custom connector by URL:

1. Open **Settings > Connectors**.
2. Click **Add custom connector**.
3. Name it `RizPay Docs` and paste the URL `https://www.rizpay.app/api/mcp`.
4. Save. No authentication is required.

Once connected, ask Claude to search the RizPay docs and it will use the `search_docs` and `get_doc` tools automatically.

### Cursor

Create `.cursor/mcp.json` in your project (or add to the existing file):

```json
{
  "mcpServers": {
    "rizpay-docs": {
      "url": "https://www.rizpay.app/api/mcp"
    }
  }
}
```

Cursor picks up the server on the next reload. You can then reference the RizPay docs from Cursor's chat and agent.

### ChatGPT

In ChatGPT, add RizPay as a custom connector by URL:

1. Open **Settings > Connectors** (custom connectors require a plan that supports them; some clients gate this behind developer mode).
2. Add a connector and paste `https://www.rizpay.app/api/mcp`.
3. No authentication is needed.

### Generic mcp.json

Most MCP clients read a config block in this shape. Adapt the key names to your client if they differ:

```json
{
  "mcpServers": {
    "rizpay-docs": {
      "type": "http",
      "url": "https://www.rizpay.app/api/mcp"
    }
  }
}
```

## Other AI-forward options

If you cannot use MCP, the docs are still easy to feed to an AI tool:

- **Copy page as Markdown** - every docs page has a "Copy page" button with a dropdown. It copies the page as clean Markdown, or opens it directly in Claude.ai, ChatGPT, or Cursor.
- **Markdown version of any page** - append `.md` to any docs URL to get raw Markdown. For example, `https://www.rizpay.app/developers/docs/getting-started/quickstart.md`.
- **[llms.txt](https://www.rizpay.app/llms.txt)** - a concise, linked index of every docs page, for AI tools that follow the [llms.txt convention](https://llmstxt.org).
- **[llms-full.txt](https://www.rizpay.app/llms-full.txt)** - the entire docs corpus concatenated into one file, ready to paste into a large-context model.
- **[OpenAPI specification](https://www.rizpay.app/api/partners/openapi)** - the machine-readable API spec, useful for generating clients or grounding an assistant in exact request and response shapes.

## Build faster: a quick example

With the MCP connected, you can let your assistant do the doc lookup for you. Try a prompt like:

> Using the RizPay docs MCP, show me how to check my account balance and then buy MTN airtime. Include the exact endpoints, headers, and request body.

Your assistant will call `search_docs` to find the relevant pages, `get_doc` to read them in full, and then write the integration against the real RizPay API - for example the [Quick Start](/developers/docs/getting-started/quickstart) flow, complete with the `external_reference` rules from [Duplicate Prevention](/developers/docs/core-concepts/duplicate-prevention).

Because the assistant is reading the live docs, it stays current with the API instead of relying on stale training data.

## Next Steps

- [Quick Start](/developers/docs/getting-started/quickstart) - Make your first API call in 5 minutes
- [Authentication](/developers/docs/getting-started/authentication) - API keys, scopes, and environments
- [Error Handling](/developers/docs/getting-started/errors) - Response format and error codes
- [API Explorer](/developers/docs/api) - Interactive API documentation