#25 Digest

This week’s roundup dives deep into tackling flaky tests head-on with smart retry strategies and CI pipelines that actually make triaging easier. You’ll also find fresh takes on optimizing resource costs during test runs, mastering full-page screenshots in tricky layouts, and even automating Playwright without letting AI run wild. Plus, there’s practical coverage on testing email flows and verifying privacy claims that you won’t want to miss.

Browser MCP token cost: 68 runs of Playwright MCP vs Chrome DevTools MCP, and a proxy that cuts snapshots by up to 96%

This article benchmarks Playwright MCP versus Chrome DevTools MCP in Claude Code and breaks down where browser-agent token costs actually come from. It also introduces a proxy that can drastically shrink large DOM snapshots, with measured tradeoffs and concrete cost reductions.

How to Make Playwright Autonomous (Without Letting an AI Agent Run Wild)

Shows how to build a tightly constrained Playwright agent loop that uses accessibility snapshots, one-action-at-a-time model decisions, and independent assertions to keep the LLM from becoming the test oracle. It also covers failure handling, trace capture, and the tradeoffs of autonomous browser testing in CI.

Full-page screenshots that are actually complete: lazy loading, sticky headers, 100vh layouts and emoji

A practical deep dive into making full-page screenshots actually reliable with Playwright/Puppeteer: handling lazy-loaded content, sticky/fixed overlays, 100vh layouts, font rendering in containers, and cookie/bot checks. It also compares real-world capture quality across three screenshot tools on a benchmark of 30 sites.

Playwright Flaky Tests: Detect Them in CI with Retries

Shows how to use Playwright retries as a flake detector in CI, then build a pipeline that captures JSON/blob reports, aggregates flaky outcomes across shards, and ranks tests by flake rate over time. It also covers a practical confirmation workflow with repeatEach, quarantine tagging, and failOnFlakyTests to keep new flakes from slipping in.

Playwright Video Recording in CI: Modes, Cost and Fixes

A practical deep dive into Playwright video recording in CI, including the subtle differences between video retention modes, how retries affect what gets recorded, and how to avoid missing or empty artifacts. It also covers cost tradeoffs, artifact upload patterns for sharded CI, and when trace or custom recording approaches are better than default video capture.

"Your file never leaves your device": how to test that claim (and the Chromium catch)

Shows how to turn a privacy claim like “your file never leaves your device” into a Playwright test that monitors network requests for accidental uploads. It also highlights a Chromium-specific blind spot where file bodies may be hidden from postDataBuffer(), and explains why a control test is essential.

Testing Email Verification with Playwright and DevMail

Shows a practical pattern for testing email verification flows in Playwright by combining browser actions with an external mailbox API, including aliasing, polling, and parsing the verification link from the message body. Useful for teams that need to validate signup, reset-password, or OTP journeys end-to-end.

Keep reading

Subscription details

You are receiving this because you are a free subscriber to Playwright Weekly.