AI Testing 17 min
AI Agents Hallucinate Domains: --allowed-origins Leaks
AI agents navigate to hallucinated domains when nobody gives them a URL. In my tests, Playwright MCP's --allowed-origins still let Chrome connect first.
Read the story Currently writing · open to consulting
Test automation that survives production.
I'm Halmurat T. — senior SDET writing about test automation at enterprise scale. Fifteen years debugging production systems at Fortune 500 telecom, insurance, retail, and financial services. The kind of scars you can't read about in documentation.
This blog is opinionated by design. Playwright vs Selenium isn't a tie. Page Object Model is overengineered. Your flaky test isn't flaky — it's broken. If that sounds useful, stick around.
ship/ StepScout — IntelliJ plugin for Cucumber step navigation →
ship/ Rally Work Items — Rally tickets inside your JetBrains IDE →
Don't miss a thing
Subscribe to get updates straight to your inbox.
No spam · Unsubscribe anytime
§02 Status board
§01 Recent writing
View all →
14 min
readOnlyHint: true, Yet Playwright's MCP Tools Write Files
Playwright 1.65.0 still marks test_run and generator_write_test readOnlyHint: true, so auto-approve MCP clients skip the prompt. Here's a checker to catch it.
16 min
Trivy and Grype Miss Chromium in Playwright Docker Images
Trivy and Grype found 0 Chromium CVEs when I scanned the Playwright Docker image. How to scan Playwright Docker images properly, and a CI gate that catches it.
18 min
Playwright Test Agents in Java: Plan, Generate, Heal
Playwright Test Agents for Java aren't planned. This MCP, CLI and Browser.bind() setup still planned, wrote and healed 25 TestNG tests in a Maven project.
15 min
Parallel Playwright MCP Agents Need Their Own Connection
Parallel agents on one Playwright MCP connection overwrite each other, and isolatedContext won't stop it. Real runs, plus the per-agent setup that went 12/12.
11 min
Agentic Testing vs Playwright Tests: Who Gets the PR Gate?
Agentic testing vs Playwright tests answer different questions. A 60-run demo plus Slack's 200-run study show which one earns the PR gate, and where agents fit.