An AI agent that designs, writes and runs your tests.
A desktop agent for your own repo. It plans, writes and runs your tests, then heals broken scripts and fixes flaky ones.
Test the checkout flow with a saved card.
- Opened localhost:3000/cart in the browser
- Wrote tests/checkout.spec.ts
tests/checkout.spec.ts
import { test, expect } from "@playwright/test"; test("pays with a saved card", async ({ page }) => { await page.goto("/cart"); await page.getByRole("button", { name: "Checkout" }).click(); await page.getByLabel("Visa ending 4242").check(); await page.getByRole("button", { name: "Pay now" }).click(); await expect(page.getByText("Order confirmed")).toBeVisible();}); pays with a saved card1.9s
Now try a card that gets declined
What the agent does
Explored /cart, /checkout, /orders
- Pay with a saved card
- Pay with a new card
- Show an error for a declined card
- Apply a discount code before paying
Approve plan
Checkout with saved and declined cards
- pays with a saved cardHappy path
- shows an error for a declined cardError
- keeps the cart after a declined cardError
test("shows an error for a declined card", async ({ page }) => { await page.goto("/checkout"); await page.getByLabel("Card number").fill(cards.declined); await page.getByRole("button", { name: "Pay now" }).click(); await expect(page.getByRole("alert")) .toHaveText("Your card was declined.");}); npx playwright test
- updates the cart total
- shows the receipt Fix: wait for the receipt heading, not a timer.
await page.goto("/cart");- await page.getByRole("button", { name: "Checkout" }).click();+ await page.getByRole("button", { name: "Continue to payment" }).click(); await expect(page.getByText("Order confirmed")).toBeVisible(); Evidence: same button on /cart, renamed “Continue to payment”.
Assertions unchanged
How it works
Open your project
Pick a folder. It learns how your tests are set up.
Ask in plain words
“Test the checkout flow.” Watch, or take over.
Keep the tests
Ordinary test files that run in CI with no AI.
Your key stays with you
Use the key from the model provider you choose and pay them directly. The key stays in your operating system's keychain.
Questions
- Do I need an account?
- No. Probeflow Desktop needs no account and no sign-in.
- Does Probeflow see my API key or my code?
- Your key stays on your device and is sent only to the model provider you choose. We never see it, and we do not store prompts or output.
- What happens when a test is flaky?
- It is tagged Flaky and can be quarantined so CI stays green. Then the agent proposes a fix.
Download
Probeflow Desktop is free. No account, and it runs on your machine.
Not available yet. Keep in touch and we will let you know when it is ready: contact@probeflow.net.