Skip to main content
By the end of this guide an AI agent can create checkouts, check payments and manage subscriptions for your store. You ask in plain words; the agent does the work. There are two ways in. You can set up both; they do not conflict.

Connect the MCP server

Claude.ai, ChatGPT, Claude Code or Cursor. You approve the connection once in the browser and paste no key.

Give a coding agent the CLI

Any agent that runs shell commands, including one in CI. It uses the agenta CLI and the SKILL.md prompt.

​
Before you start

You do not need an account up front. When you approve the connection or run agenta login, you sign in, and a new email creates your account.

​
Path A: connect the MCP server

1

Open the setup dialog

In the dashboard, open Developers → API keys in the sidebar and click Connect an AI agent. The Set up your AI agent dialog asks “Where does your agent run?” Pick your client: Claude Code, Cursor, ChatGPT, Claude.ai, or Other. The dialog shows the steps for that client.
The Set up your AI agent dialog in Settings. The Claude Code tab is selected next to Cursor, ChatGPT, Claude.ai and Other. Step 1, Install the plugin, shows two claude plugins commands with copy buttons. Step 2, Approve it here, says to type /mcp, choose agentaos, then Authenticate, and pick Test or Live on the AgentaOS page.

The Set up your AI agent dialog with Claude Code picked: install the plugin, then approve it in AgentaOS.

The Set up your AI agent dialog with the Claude.ai tab selected. Step 1, Add the connector, says: In Claude, Customize, Connectors, Add custom connector, and paste this URL, followed by https://mcp.agentaos.ai/mcp with a copy button. Step 2, Approve it here, says a page opens on AgentaOS where you pick Test or Live and press Approve.

The same dialog with Claude.ai picked: add the connector, then approve it in AgentaOS.

2

Connect your agent

Follow MCP server for your client. For Claude.ai or ChatGPT, follow Connect Claude.ai or ChatGPT. Pick Test mode on the approval page while you try things out.
3

Try a prompt

Ask your agent:
Create a checkout for 50 EUR for my consulting invoice.
The agent calls agenta_pay_checkout and answers with the link to share, in a sentence like this one: “Checkout for €50.00 created. Share https://… with your customer.”
4

Manage what you sold

Ask it directly:
List my last 5 checkouts.
Cancel the subscription for ana@example.com at the end of the current period.
The agent calls the matching tools, such as agenta_pay_list, agenta_subscriptions_list and agenta_subscriptions_cancel, and reports back in the chat.
MCP server lists every tool your agent can use.

​
Path B: a coding agent with the CLI

1

Install the CLI and log in

Install the agenta CLI with the command on Install the CLI, then sign in with agenta login. See Login.agenta login is the one step that needs you: it opens a browser for sign-in. Every command an agent runs after that is non-interactive.
2

Give your agent the skill

Tell your coding agent to fetch and follow the skill prompt:
Read https://agentaos.ai/SKILL.md and set up AgentaOS for me
In Claude Code, you can install the plugin instead. See AI agent setup.It teaches the agent the agenta commands and their flags. It also tells the agent to run agenta status --json first and to pass --json to every command, so the output is JSON the agent can read.
3

Try a prompt

Ask it something like:
Create a checkout for 50 EUR for my consulting invoice.
The agent runs agenta status --json to confirm payment tools are ready, then agenta pay checkout -a 50 --json, and hands you back the checkoutUrl to share with your client.
An agent connected in test mode, or running on a test key, sees and creates test data only. It cannot move real money.

​
Verify it worked

Ask the agent:
What’s my AgentaOS status?
On the MCP path, the agent calls agenta_status_get. Before you go live, the answer starts with “Test payments work now.” On the CLI path, the agent runs agenta status --json and reports paymentTools.ready: true, along with your business.

​
Next steps

MCP server

Setup for each client, and every tool your agent can use.

CLI install

The full agenta CLI reference, for scripts outside an AI agent too.

Receive webhooks in Express

Checkouts an agent creates still need your server to fulfill them on completion.