Agent access

AI agents are first-class users. Everything a person can do, an agent can do through the API or the MCP server, with the same account, the same allowance and the same price: run a check on demand, add a store to the schedule, read the last 48 results, change where alerts go.

Discover the product

• https://checkoutbeat.com/llms.txt: a plain-text summary written for language models.
• https://checkoutbeat.com/.well-known/product.json: a machine-readable description of the feature, its inputs and outputs, prices and endpoints.
• https://checkoutbeat.com/api/v1/openapi.json: the OpenAPI 3.1 document, generated from the same schemas the server validates with.

Call the API

Authenticate with an API key and send X-Usage-Source: agent so the account's usage meter attributes the checks to agents rather than to plain API calls. The header is advisory: it changes reporting, never the price.

POST /api/v1/store-check
Authorization: Bearer checkoutbeat_live_…
X-Usage-Source: agent
Idempotency-Key: 7d2c…
Content-Type: application/json

{"storeUrl": "https://your-store.com", "flowType": "auto"}

The call waits for the browser walk (10–40 seconds on a healthy store, a minute at most) and answers with result.steps, result.checks and result.summary, e.g. "checked at 14:05 UTC, step Checkout page failed with HTTP 500". When the result is not ready in time, or with "wait": false, it answers 202 with a job to poll at GET /api/v1/jobs/{jobId}.

Reading is free: GET /api/v1/stores lists the monitored stores with status and 30-day pass rate; GET /api/v1/stores/{id} adds the last 48 results, the 48-slot timeline and the incidents.

MCP server

The product ships an MCP server with two tools. It runs on the agent's side and calls the API with the agent's key:

{
  "mcpServers": {
    "checkoutbeat": {
      "command": "npm",
      "args": ["run", "mcp", "--prefix", "/path/to/checkoutbeat"],
      "env": {
        "PRODUCT_API_URL": "https://checkoutbeat.com",
        "PRODUCT_API_KEY": "checkoutbeat_live_…"
      }
    }
  }
}

• store-check: run a check on demand. Same input as the API; metered as one on-demand check with source agent.
• store-results: read a monitored store's last 48 results by store id or store URL. Not metered.

A check needs a real browser, which runs on CheckoutBeat's workers; the MCP server never launches one on the agent's machine.

Prices for agents

Accounts on a plan use the plan's monthly allowance of on-demand checks. When it is used up, agent and API checks from an account with a card on file are billed at $0.25 per check. Free accounts get 1 on-demand check per month and HTTP 402 afterwards, with a link to the pricing page.

Rules for agents

• Respect Retry-After on 429 responses.
• Send an Idempotency-Key with every check request and reuse it on retries.
• Read error.message; it is written to be acted on. A 400 saying the store could not be walked automatically means the store needs a product URL or a recorded buy path, not another attempt.
• Do not create more API keys than you need; key management is deliberately session-only.
• The walk stops at the checkout page. There is no way to make it pay, and there is nothing to gain by asking.