Monitor a third-party flow

Some flows you depend on leave your app: Stripe checkout, a Calendly embed, an OAuth partner. You do not own the other side, but a break is still your outage. This recipe runs a short flow against the real integration on a schedule and pages you on failure.

The test

Keep it to the boundary: your app, into the vendor surface, far enough to prove the integration is alive, and back:

tests/monitors/checkout-stripe.test.yaml

fileType: momentic/test/v2
id: stripe-checkout-monitor
url: https://app.example.com/pricing
labels: [monitor]
before:
  - module: ../modules/log-in.module.yaml
steps:
  - click: Upgrade plan
  - assert: A Stripe checkout form loads and shows the plan price
  - type:
      text: 4242 4242 4242 4242
      into: the Card number field
  - type:
      text: "12/34"
      into: the Expiry field
  - type:
      text: "424"
      into: the CVC field
  - assert: The Pay button is enabled

Stop before the irreversible action: assert the form is live, not that a charge posts. When the vendor offers a test mode (Stripe test keys, a sandbox tenant), point the monitor at that instead and run the full path.

The workflow

The job installs momentic from devDependencies, so your lockfile pins the CLI version. Add it once with npm install -D momentic.

.github/workflows/monitor.yml

name: Third-party monitors

on:
  schedule:
    - cron: "*/15 * * * *" # every 15 minutes
  workflow_dispatch:

jobs:
  monitor:
    runs-on: ubuntu-latest
    timeout-minutes: 10
    env:
      MOMENTIC_API_KEY: ${{ secrets.MOMENTIC_API_KEY }}
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: 22.12.0
      - run: npm install
      - run: npx momentic install-browsers chromium

      - name: Run monitors
        run: npx momentic run --labels monitor -y --upload-results

      - name: Alert on failure
        if: failure()
        uses: slackapi/slack-github-action@v2
        with:
          webhook: ${{ secrets.SLACK_WEBHOOK_URL }}
          webhook-type: incoming-webhook
          payload: |
            text: ":rotating_light: Third-party monitor failed: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"

Scheduled runs get the same secrets as other events. GitHub skips workflow runs on forks and on repos whose Actions were disabled. Confirm the job is firing (the workflow_dispatch trigger is also your debug path).

Third-party fragility rules

Their outage is your alert, not your test’s bug. A monitor failing because Stripe is down is the point. Route monitor failures to an on-call channel, separate from your PR-gate failures.
Vendor UI drifts. Payment iframes change markup. Write natural-language targets (“the Card number field”) so locator auto-heal absorbs their redesigns.
Do not retry into a charge. --retries is safe on read-only monitors; keep it off anything that mutates real state. Prefer vendor test modes.
Rate limits are real. Fifteen minutes is a reasonable floor; a flow that takes minutes to run should not run every minute. If you need tighter detection, monitor the vendor’s status endpoint with a plain HTTP check and keep Momentic for the end-to-end signal.

Overview

Migrate to Momentic

Frameworks

Use cases

Triage

AI actions

Authentication

Page interactions

Test data & variants

Visual testing

Reporting

Upgrading

Monitor a third-party flow

The test

The workflow

Third-party fragility rules

​The test

​The workflow

​Third-party fragility rules

​Related

The test

The workflow

Third-party fragility rules

Related