Coming soon

Catch broken signup & checkout
before your users do.

AI smoke tests that run on every deploy and on a schedule — so your vibe-coded app stays reliable. Try a free test now — get notified when we launch.

No install, 2 minutes. Works best for modern web apps (React, Next, etc.). Captcha/MFA flows need extra setup.

Built for indie devs shipping daily.

Stop learning about regressions from angry users, refund emails, or a dip in conversions.

< 5 min

to first result

0

selectors to write

0

config files

Used by indie builders, side projects, and small SaaS teams.

The real problem

Shipping is easy now.
Not breaking things isn't.

AI tooling makes it trivial to push changes fast. It also makes it trivial to accidentally break:

Signup flows
Login & session handling
Checkout & payments
Forms and onboarding steps
Critical navigation paths

Autosmoke runs your critical flows like a real user and alerts you the moment something breaks — ideally before a single customer hits it.

How it works

Three steps. That's it.

No SDKs, no test runners, no browser drivers. Just results.

Step 01

Point Autosmoke at your site

Paste a URL — staging or production. No config files, no setup scripts.

Step 02

Describe a flow in plain English

"Sign up with email → confirm dashboard loads → log out." That's your entire test.

Step 03

Autosmoke runs it automatically

On every deploy (GitHub Actions / CI), on a schedule (every 5/15/60 min), or on demand with one click.

What you get back

A report you can actually use.

Not just pass/fail. Every run comes with a full timeline, annotated screenshots, session video, and a shareable link you can drop in Slack or a PR comment.

Fewer false alarms — Autosmoke focuses on smoke checks for critical paths, not brittle UI micromanagement.

Step-by-step timelineScreenshots at each stepSession video replayError summary in plain EnglishShareable link for teammates

SMOKE TEST REPORT

your-app.com/checkout

Passed

Timeline

0.0s
Navigate to /checkout
1.2s
Fill email and payment form
3.1s
Click 'Place Order'
4.2s
Confirm success page loaded
4.2s total|

Why it's different

Intent-based, not selector-based.

Selectors change, timing shifts, and your tests fail even when your app is fine. Then you spend your weekend “fixing tests,” not shipping. Autosmoke follows the flow like a user would.

cart.spec.tsFragile
1import { test, expect } from "@playwright/test";
2
3test("user can add item to cart", async ({ page }) => {
4 await page.goto("https://store.example.com");
5 await page.locator(
6 'div.nav__wrapper > ul > li:nth-child(3) > a[data-testid="shop-link"]'
7 ).click();
8 await page.waitForSelector(
9 ".product-grid__container .product-card:first-child"
10 );
11 await page.locator(".product-card:first-child .btn--add-to-cart").click();
12 const cartBadge = page.locator("span.cart-icon__badge");
13 await expect(cartBadge).toHaveText("1");
14});
add-to-cart.smokeResilient
1Navigate to the Shop page
2Click 'Add to Cart' on the first product
3Verify the cart badge shows 1 item
All 3 steps passed in 4.2s
Finds the right button/link like a user would
Waits for the page to actually be ready
Adapts to small UI changes automatically
Produces a human-readable diagnosis when it fails

Less maintenance. More shipping.

Integrations

Runs where you already ship

GitHub Actions

Block or warn on broken flows before merge/deploy.

# .github/workflows/smoke.yml
- name: Run smoke tests
uses: autosmoke/action@v1
with:
url: ${{ vars.STAGING_URL }}
api-key: ${{ secrets.AUTOSMOKE_KEY }}

Copy-paste YAML. Works in minutes.

Docs

Scheduled Monitoring

Get alerted if production breaks at 2am.

Every 5 / 15 / 60 min

Run on a schedule. No babysitting required.

Notify via:

EmailSlackWebhook

Questions.

(We have answers.)