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

# Build with AI agents

> Don't integrate by hand. Point your AI agent at AgentaOS and it installs the CLI, creates a product, and hands you a live checkout link.

You build with agents. So don't read an SDK reference and translate it into code yourself. **Hand AgentaOS to your agent and let it do the integration.**

<Card title="Paste this into Claude Code, Cursor, or any coding agent" icon="robot" href="https://agentaos.ai/SKILL.md">
  ```text theme={null}
  Read https://agentaos.ai/SKILL.md and set up AgentaOS for me:
  install the CLI, log me in, create a test product, and give me the checkout link.
  ```
</Card>

It reads [`SKILL.md`](https://agentaos.ai/SKILL.md) (the exact commands, flags, and response shapes), installs the `agenta` CLI, signs you in, creates a product in test mode, and hands you a shareable checkout URL. You approve the browser sign-in once; everything after that is non-interactive.

## Connect the MCP server

For Claude Desktop, Cursor, or Windsurf chat, connect the MCP server and your agent can run the store: create checkouts, read customers, and manage subscriptions.

```json theme={null}
{
  "mcpServers": {
    "agenta": {
      "command": "npx",
      "args": ["-y", "agentaos"],
      "env": { "AGENTAOS_GATEWAY_KEY": "sk_test_your-gateway-key" }
    }
  }
}
```

Test keys only ever see test data, so an agent on a `sk_test_` key can build and break things safely. See [MCP server](/mcp/setup) for Cursor and Windsurf configs and the full tool list.

## Feed the docs to your model

The whole documentation is available as plain markdown for LLMs. Point your model at it or paste it into context:

<CardGroup cols={2}>
  <Card title="llms.txt" icon="list" href="/llms.txt">
    A compact index of every page, for grounding a model.
  </Card>

  <Card title="llms-full.txt" icon="book" href="/llms-full.txt">
    The entire docs as one markdown file. Everything, in one paste.
  </Card>
</CardGroup>

## What your agent can run

| Tool                             | What it does                                              |
| -------------------------------- | --------------------------------------------------------- |
| `agenta_pay_create_checkout`     | Create a checkout and get a shareable `checkoutUrl`.      |
| `agenta_pay_get_checkout`        | Check whether a payment completed.                        |
| `agenta_pay_list_checkouts`      | List checkouts by status.                                 |
| `agenta_pay_list_subscriptions`  | List subscriptions: status, amount, interval, subscriber. |
| `agenta_pay_cancel_subscription` | Cancel, at period end or immediately.                     |
| `agenta_pay_list_customers`      | Everyone who paid you: email, name, country.              |
| `agenta_pay_send_receipt`        | Re-send a receipt email.                                  |

## Next steps

<CardGroup cols={2}>
  <Card title="Run your store from an agent" icon="wand-magic-sparkles" href="/guides/ai-agent">
    The full walkthrough: coding agent with the CLI, or a chat agent with MCP.
  </Card>

  <Card title="MCP server" icon="server" href="/mcp/setup">
    The payment tools, and Claude / Cursor / Windsurf setup.
  </Card>

  <Card title="Take a payment yourself" icon="rocket" href="/guides/accept-your-first-payment">
    Prefer to do it by hand first? Create a product and get paid.
  </Card>
</CardGroup>