npx @kifas/mcp@latest

Take QA automation off your to-do list.

Connect your GitHub repo. Kifas agents read the code, write the tests, and run them on real browsers, emulators, and the Device Farm. Or vibe-test in plain language: describe a flow and it becomes a running test.

Start freeRead the MCP docs

1,000 credits ($10), no card, 14 days. Live now. Not a waitlist.

acme/storefrontmain65 MCP tools
SCANNING REPOkifas.scan
  • P0Checkout with a saved cardcheckout.spec.ts
  • P1Sign up, verify email, reach the dashboardsignup.spec.ts
  • P2Invite a teammate and accept the inviteinvite.spec.ts
Reading routes, components, and the API calls behind them.
Works with the tools you already use
PlaywrightAppiumClaudeCursorWindsurfGitHubJira
See it work

A testing agent built for real devices.

It writes the tests, runs them on phones and browsers, and tells you what actually broke.

01

The first mobile-optimized AI testing harness

Built for phones and browsers. It knows the platform it's on and taps, swipes, long-presses and uses hardware keys the way that platform expects.

02

Guide it while it builds

Watch every step run live on the device and reply to the agent when it goes the wrong way. Saved tests replay without the AI, so they run faster and cost less.

03

Deterministic pixel checks + AI checks

Exact pixel comparison when every pixel matters, an AI check when you only care that the screen is right. Pixel snapshots are free.

04

You decide what heals and what fails

Pick how strict each suite is. A renamed button heals itself, while a changed expectation always waits for you.

05

Emulators and real iOS and Android devices

Start on the Android emulator, then run on real Android phones in our Device Farm, and on real iPhones through your Appium suite. About 40% less than teams pay BrowserStack.

06

Checks on every pull request

Results land as GitHub checks before you merge. Every web test reads as standard Playwright code.

Start free with 1,000 credits
Google Pixel 8 · Android 15
tapswipe uplong presstypebackhomelaunch app
✓ launch app com.acme.shop
✓ tap "Sign in"
✓ type email into the native keyboard
✓ wait for "Your orders"
→ swipe left on order #2 to reveal "Cancel"
· tap "Cancel"
· press back
Taps, swipes and hardware keys
New: visual testing

Catch visual bugs before your users do.

Deterministic pixel checks and plain-language AI checks, with a one-click review queue. Pixel snapshots are free.

Playwright and AppiumBrowsers and real devicesThe Percy alternative
See visual testing
baselinenew
Switching from BrowserStack?

Real devices for about 40% less. Plus an agent that tells flaky from broken.

Teams moving from BrowserStack typically save around 40%. Based on quotes we've given to teams currently on BrowserStack. Your saving depends on your devices and concurrency; send us your quote and we'll show you the number.

KifasBrowserStack
Real iPhones and Android phones✓✓
How you payCredits for what you runPlans priced per dedicated parallel device
Visual snapshotsPixel snapshots freePercy, priced by screenshots

How switching works

Try it on your own GitHub repo.
1,000 credits ($10), no card, 14 days.
Start free →
MCP-first, not MCP-bolted-on

Built MCP-first, not bolted on.

Most tools added an MCP server after the fact. Kifas was built around it. The same 73 tools the autonomous loop uses are the ones you get when you connect your own agent: spawn a device, navigate, assert, save state, tear down. Whatever your agent can reason about, it can now test.

~/acme-web · claude-code
// Install Kifas's MCP server
$ npx @kifas/mcp@latest --install cursor
→ Installed 73 tools. Authenticated as maya@acme.dev
 
claude > "Test our checkout with a new Gmail user, apply promo GUMBO20, verify the total."
 
→ Kifas: workflow.create("checkout-promo")
→ Kifas: workflow.add_node(type: "auth", primitive: "gmail")
→ Kifas: workflow.add_node(type: "action", ai_explore: true)
→ Kifas: workflow.run() · 14 nodes · 3.1s start · passed
 
// Compiled to /tests/checkout-promo.spec.ts
Already have a suite?

Import it. Kifas makes it fast.

Point Kifas at the suite you already run. For each run it flags the steps that never needed a browser and suggests a faster API check, and the rest stays on real browsers.

01
Import runs your suite as-is

Point Kifas at your Playwright or Appium suite. It runs first exactly as it does in your CI, so nothing changes until you decide it should.

02
It watches the real traffic

Kifas captures every network request and response behind each step. Auth tokens are managed in the platform, so it can verify backend calls directly.

03
Each step gets classified

Safe to convert to an API-level check, or genuinely browser-dependent. Visual checks and client-side logic stay in the browser. Everything else drops a layer.

04
You approve the diff

The proposed conversion lands as a reviewable diff. Nothing converts silently. Approve what you trust and keep the rest running in the browser.

Try it on your own suite. The free tier is 1,000 credits ($10), no card. Migration guides

Start free. Scale when it pays off.

1,000 credits ($10), no card, 14 days. Pay for what you run, not for seats. Pro is $199/mo.

Free
$0
$10 in credits (1,000 credits) · no card · 14 days
  • Cloud browsers, emulators, Device Farm
  • MCP server and 65 tools
  • Community support
Start free
Pro
$199/mo
The autonomous loop, self-healing, and the full device range.
  • GitHub-connected autonomous loop
  • Self-healing with break thresholds
  • Auth Catalog · vibe testing · hybrid engine
Choose Pro
Custom
Let's talk
For teams testing on real devices at scale.
~40% less
than teams pay BrowserStack for real devices*
  • Everything in Pro
  • Dedicated real devices in our Device Farm
  • Highest concurrency
  • SSO, audit logs and a DPA
  • Premium support and onboarding
Talk to us

See all plans →

Point Kifas at your GitHub today.

1,000 credits ($10), no card. Connect a repo, or paste npx @kifas/mcp@latest and let the agent drive.

Start freeGet a quote