Build with your coding agent · WearCheckout Docs

Let your coding agent (Claude Code, Cursor and others) add WearCheckout to your app and test it, on your computer. It builds and tests. Anything with real money, and going live, stays with you.

What your agent can do

Create apps, and change them before they're live.
Set a price before you go live. Buyers pay it only after you do.
Add our library to your app's Connect IQ project.
Run test purchases and test refunds (test mode: no card, no money) until all 5 checks pass. See Testing.
Read sales, watches and buyer feedback, with a key you allowed to read live data.

What only you can do

Going live, start to finish. Live prices and live app settings. Connecting a payment processor. Refunds, disputes and payouts. Moving a paid purchase. Replying to buyers. Billing, your plan, your team and agent keys. There's no tool for any of these: your agent hands you a link.

1. Make a key

In your dashboard, open Settings, Agents and choose New key. Name it after the agent and the computer ("Claude Code on my laptop"), and pick what it can do:

Kind	Starts with	Can
Test mode only	`wck_test_`	Build and test. Sees no live data.
Test mode, and reads live data	`wck_read_`	Also reads live sales, watches and feedback. Changes nothing live.

We show the key once. Treat it like a password.

2. Connect your agent

Claude Code:

Terminal

claude mcp add wearcheckout -e WEARCHECKOUT_KEY=YOUR_KEY -- npx -y https://wearcheckout.com/mcp/wearcheckout-mcp.tgz

On Windows, Claude Code starts it through cmd /c:

Terminal

claude mcp add wearcheckout -e WEARCHECKOUT_KEY=YOUR_KEY -- cmd /c npx -y https://wearcheckout.com/mcp/wearcheckout-mcp.tgz

Cursor, and other agents that read an MCP config:

mcp.json

{
  "mcpServers": {
    "wearcheckout": {
      "command": "npx",
      "args": ["-y", "https://wearcheckout.com/mcp/wearcheckout-mcp.tgz"],
      "env": { "WEARCHECKOUT_KEY": "YOUR_KEY" }
    }
  }
}

It needs Node 18 or later.

3. Ask it

"Add WearCheckout payments to this app." It finds or creates the app, sets the price, adds the library, builds the app and opens it in the simulator, then runs the test purchase and refund. When all 5 checks pass, it gives you the link to go live. See Going live.

What it sees

Watch codes and models, never buyer emails or names. Buyer feedback reaches it marked as written by strangers, so it can't steer your agent.

Its tools

Tool	Does
`get_account`	What this key may do
`list_apps`, `get_app`	Your apps
`create_app`	A new draft app
`update_app`, `set_price`	Change an app that isn't live
`get_install`	What to add to the project
`get_test`, `run_test_purchase`, `run_test_refund`	The 5 test checks
`list_sales`, `get_sale`, `list_watches`, `get_watch`, `get_stats`	Read (live with a read key)
`list_feedback`	Buyer messages
`go_live_link`	The link for you to go live

Stop an agent

In Settings, Agents, choose Revoke. It stops at once. Everything an agent did is listed under Recent agent activity, including what it asked for and was refused.