> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rampnow.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Build with AI

> Connect Rampnow's docs to Claude, Cursor, ChatGPT and other AI tools so your agent builds against current docs

Rampnow's docs are published in formats AI tools can read directly. Connect your coding agent to them, and it works from the current API instead of guessing from its training data or from pages you paste in.

<CardGroup cols={2}>
  <Card title="MCP server" icon="plug" href="#mcp-server">
    Your agent searches and reads the docs while it works
  </Card>

  <Card title="llms.txt" icon="file-lines" href="#llms-txt">
    An index of every page, or the whole site in one file
  </Card>

  <Card title="Markdown export" icon="markdown" href="#markdown-export">
    Any page as clean Markdown
  </Card>

  <Card title="Agent skill" icon="wand-magic-sparkles" href="#agent-skill">
    A short brief that tells agents how to integrate Rampnow
  </Card>
</CardGroup>

## MCP server

The docs run a hosted [Model Context Protocol](https://modelcontextprotocol.io) server at:

```text theme={null}
https://docs.rampnow.io/mcp
```

It gives your agent two read-only tools: one searches the docs, the other reads full pages and the OpenAPI spec. It needs no authentication.

<Tabs>
  <Tab title="Claude Code">
    ```bash theme={null}
    claude mcp add --transport http rampnow-docs https://docs.rampnow.io/mcp
    ```
  </Tab>

  <Tab title="Claude">
    In Claude, open **Settings → Connectors**, choose **Add custom connector**, name it `Rampnow Docs` and enter `https://docs.rampnow.io/mcp` as the URL.
  </Tab>

  <Tab title="Cursor">
    Add the server to `~/.cursor/mcp.json` (all projects) or `.cursor/mcp.json` (one project):

    ```json theme={null}
    {
      "mcpServers": {
        "rampnow-docs": {
          "url": "https://docs.rampnow.io/mcp"
        }
      }
    }
    ```
  </Tab>

  <Tab title="VS Code">
    Add the server to `.vscode/mcp.json`:

    ```json theme={null}
    {
      "servers": {
        "rampnow-docs": {
          "type": "http",
          "url": "https://docs.rampnow.io/mcp"
        }
      }
    }
    ```
  </Tab>

  <Tab title="ChatGPT">
    In ChatGPT, open **Settings → Apps & Connectors**, create a connector and enter `https://docs.rampnow.io/mcp` as the server URL. Custom connectors need developer mode turned on.
  </Tab>
</Tabs>

Clients that support discovery can find the server at `https://docs.rampnow.io/.well-known/mcp`.

You can also add the server from any docs page: open the page menu at the top of the page and choose **Connect to Cursor**, **Connect to VS Code** or **Copy MCP server URL**.

<Note>
  The MCP server searches the documentation only. It doesn't call the Rampnow API, see your orders or use your API keys.
</Note>

## llms.txt

Two files follow the [llms.txt](https://llmstxt.org) convention:

| File | Contents | Use it when |
| - | - | - |
| [`/llms.txt`](https://docs.rampnow.io/llms.txt) | Every page, with its title, description and Markdown link | Your agent fetches the pages it needs |
| [`/llms-full.txt`](https://docs.rampnow.io/llms-full.txt) | The full text of every page in one file | You give a model the whole site in one go |

To ask questions about the full docs from your own code, send `llms-full.txt` as a cached system prompt. Follow-up questions then reuse the cache, which costs less and responds faster:

```python Python theme={null}
import urllib.request

import anthropic

with urllib.request.urlopen("https://docs.rampnow.io/llms-full.txt") as res:
    docs = res.read().decode("utf-8")

client = anthropic.Anthropic()

response = client.beta.messages.create(
    model="claude-opus-5-5",
    max_tokens=16000,
    betas=["server-side-fallback-2026-07-01"],
    fallbacks="default",
    system=[
        {"type": "text", "text": "Answer using the Rampnow documentation below."},
        {"type": "text", "text": docs, "cache_control": {"type": "ephemeral"}},
    ],
    messages=[
        {"role": "user", "content": "How do I sign a request to the quote endpoint?"}
    ],
)

if response.stop_reason == "refusal":
    print("The request was declined.")
else:
    print("".join(block.text for block in response.content if block.type == "text"))
```

## Markdown export

Every page is also available as Markdown. Add `.md` to its URL:

* [`/get-started/intent.md`](https://docs.rampnow.io/get-started/intent.md)
* [`/api-reference/authentication.md`](https://docs.rampnow.io/api-reference/authentication.md)
* [`/api-reference/webhook.md`](https://docs.rampnow.io/api-reference/webhook.md)

A request for the normal page URL with an `Accept: text/markdown` header returns the same Markdown:

```bash theme={null}
curl -H "Accept: text/markdown" https://docs.rampnow.io/api-reference/authentication
```

To get a page without leaving the site, open the page menu at the top of any page. From there you can copy the page as Markdown or open it in ChatGPT, Claude or Perplexity.

## Agent skill

[`/skill.md`](https://docs.rampnow.io/skill.md) is a short brief for coding agents, in the [Agent Skills](https://agentskills.io) format. It covers which product to use, authentication, the order flow, statuses and common mistakes. It links to the full pages for everything else.

Install it into your agent with:

```bash theme={null}
npx skills add https://docs.rampnow.io
```

## Recommended setup

<Steps>
  <Step title="Connect the MCP server">
    Add the [MCP server](#mcp-server) to your editor or agent so it can look things up as it works.
  </Step>

  <Step title="Tell your agent which product you're building">
    Rampnow has two integrations that don't mix. Start your agent on the right quickstart:

    * **Widget or SDK** (no backend, public key only): [Widget](/get-started/widget) or [SDK](/get-started/sdk)
    * **Intent API** (your backend creates signed orders): [Intent API](/get-started/intent)
  </Step>

  <Step title="Give it the full docs for one-off tasks">
    For a single large task, such as writing a whole integration, put [`llms-full.txt`](#llms-txt) in context with prompt caching.
  </Step>

  <Step title="Check what it builds">
    Check generated requests against the [API reference](/api-reference/overview). Test the signing code against the [test vector](/api-reference/authentication#test-vector). Run the [sandbox test plan](/api-reference/sandbox-testing#suggested-test-run) before going live.
  </Step>
</Steps>

<Warning>
  Never paste your API secret into a prompt or an agent's context. Keep it in your server's environment, and have your agent write code that reads it from there.
</Warning>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.