Open Source CLI — Free Forever

Stop guessing.
DeFlaky your tests.

Detect flaky tests in seconds with AI-powered root cause analysis. BYOK — bring your own AI key. Push results to the dashboard to track trends, get alerts, and fix reliability issues before they tank your pipeline.

$npm install -g deflaky-cli
Terminal

$npx deflaky-cli --command "npx playwright test" --runs 5


Running test suite... (5 iterations)


  42 tests stable (100% pass rate)

   3 tests FLAKY

login.spec.ts > "should redirect" — 3/5 passed

cart.spec.ts > "update quantity" — 4/5 passed

search.spec.ts > "show suggestions" — 2/5 passed

   1 test consistently failing


FlakeScore: 93.5 / 100

$ _

See exactly what's flaky

DeFlaky generates a detailed report after every run — FlakeScore, per-test pass rates, and AI-powered root cause analysis.

df

DeFlaky Test Report

Run ID: d7f42a · Apr 8, 2026, 2:15 PM · Playwright

5 iterations complete

FlakeScore

0%

Total Duration

0s

Total Tests

0

Stable

0

Flaky

0

All (45)Stable (42)Flaky (3)Failed (0)
StatusTest NameResultsPass Rate
flakyshould complete checkout flow3✓ 2✗
60%
flakyshould handle OAuth redirect4✓ 1✗
80%
flakyshould load dashboard charts4✓ 1✗
80%
passedshould render login page5✓ 0✗
100%
passedshould create new project5✓ 0✗
100%

AI Root Cause Analysis

Checkout flow flakiness caused by race condition in payment confirmation — Stripe webhook response time varies 200ms-3s. Suggested fix: add explicit wait with 5s timeout.

Works with every test framework

PlaywrightSeleniumCypressJestPytestMochaJUnitTestNGVitest

Everything you need to kill flaky tests

From detection to resolution, DeFlaky gives you full visibility into your test suite's reliability.

Instant Detection

Run your suite N times with a single command. DeFlaky identifies flaky tests by comparing results across runs — no config required.

📊

FlakeScore Dashboard

Track your test suite's reliability score over time. See which tests are getting worse and which are improving.

🔔

Slack & Email Alerts

Get notified the moment a new flaky test appears. Stop finding out about flaky tests from angry developers.

🛠

Framework Agnostic

Works with any test runner that outputs JUnit XML or JSON reports. Playwright, Selenium, Cypress, Jest, Pytest — all supported.

🚀

CI/CD Ready

Add DeFlaky to your GitHub Actions, Jenkins, or GitLab CI pipeline. Fail builds when flakiness exceeds your threshold.

👥

Team Collaboration

Share flakiness reports with your team. Assign owners to flaky tests. Track who fixed what and when.

🤖

AI Root Cause Analysis

AI analyzes your stack traces and tells you exactly why a test failed — infrastructure, app bug, test code, or flaky.

📚

AI Failure Categorization

Automatically classify every failure. No more manual triage. Works with any LLM provider you choose.

AI-Powered Analysis

DeFlaky uses AI to analyze your test failures, identify root causes, and suggest fixes — automatically.

AI Root Cause Analysis
Test: flaky — random number check
Category: FLAKY(87% confidence)

Root Cause:

Test relies on Math.random() producing non-deterministic values. Each run generates a different number, causing intermittent assertion failures.

Suggested Fix:

jest.spyOn(Math, 'random')

.mockReturnValue(0.75);

Bring Your Own Key

Use any LLM provider you prefer. Your API key never leaves your browser. We never store or log it.

Anthropic (Claude)

Best for code analysis

OpenAI (GPT-4o)

Great all-around

Groq

Ultra-fast, free tier

OpenRouter

100+ models, one API

Ollama

Run locally, 100% private

Three steps to reliable tests

01

Install the CLI

$ npm install -g deflaky-cli
02

Run detection

$ deflaky --command "npx playwright test" --runs 5
03

Push to dashboard

$ deflaky --command "npx playwright test" --runs 5 --push --token YOUR_TOKEN

Simple, transparent pricing

100% free during launch. CLI is open source forever.

CLI

$0forever
  • Unlimited local runs
  • Terminal reports
  • JUnit XML & JSON support
  • All frameworks supported
  • Open source (MIT)
Install Free

Dashboard

$0free during launch
  • Everything in CLI +
  • Unlimited projects
  • 90-day history
  • FlakeScore trends
  • Email & Slack alerts
Get Free Access

Pro

$19/month
  • Everything in Dashboard +
  • AI Root Cause Analysis
  • AI Failure Categorization
  • Unlimited history
  • Priority support
Start Free Trial

Your tests aren't flaky.
They're unmonitored.

Join QA engineers who ship with confidence.