TurnSignal blog
Practical guides for Playwright teams: flaky tests, test reporting in CI, reviewing AI-written tests and pull request failures.
October 31, 2026 · 5 min read
How npx playwright merge-reports works: blob reports from each shard, one HTML, JUnit or JSON report, a merge config, and the errors you hit on the way.
October 31, 2026 · 5 min read
Split a Playwright suite across CI jobs with --shard on GitLab CI, Azure Pipelines, CircleCI and Jenkins, balance the shards, and merge one report.
October 31, 2026 · 5 min read
Playwright's JUnit reporter: output file, every 1.63 option, what the XML contains, and why flaky tests look like passes unless you set includeRetries.
October 31, 2026 · 5 min read
How --last-failed picks tests from test-results/.last-run.json, why it runs the whole suite in a fresh CI job, and how to use it safely with shards.
October 31, 2026 · 5 min read
What Playwright's error-context.md file contains (error, page snapshot, test source), where it is written, how coding agents use it, and what to keep out of it.
October 31, 2026 · 5 min read
Why npx playwright test says "No tests found": testDir, file names, .gitignore, filters, --last-failed, and when Playwright skips the check.
October 16, 2026 · 4 min read
Why a Playwright CI step ends with exit code 137, how to confirm it was memory, what to change first, and how to keep the results the killed run never reported.
October 12, 2026 · 4 min read
Playwright reports describe one run. How to keep test history across CI runs: JSON reports, a small script to find flaky and failing tests, and its limits.
October 6, 2026 · 7 min read
How npx playwright show-report works, why localhost:9323 does not open, opening a report zip from CI, host and port, and sharing the report.
October 6, 2026 · 5 min read
What "Error: worker process exited unexpectedly (code=…, signal=…)" means in Playwright, how to read code and signal, and how to fix memory and crash causes.
October 6, 2026 · 5 min read
Why Playwright throws "Target page, context or browser has been closed" in locator.click, page.goto or browser.newContext, and how to fix each cause.
October 6, 2026 · 4 min read
Playwright's retryStrategy: 'isolated' retries failed tests after all others, one at a time. When it helps, what it costs, and how to keep flaky tests visible.
October 6, 2026 · 5 min read
A Playwright trace records network headers, bodies, console output and DOM snapshots. What can leak into trace.zip, and how to keep tokens out of CI artifacts.
October 2, 2026 · 5 min read
How to run Playwright tests in GitHub Actions with sharding, retries, artifacts, and PR comments. Complete workflow file with explanations.
September 28, 2026 · 7 min read
We read 11,618 GitHub Actions runs from 79 open-source Playwright projects: 42% failed, 1 in 13 was re-run, and many failing tests hit unrelated branches.
September 28, 2026 · 4 min read
Re-running failed jobs until CI is green hides real bugs and burns CI time. Four questions that tell a flaky Playwright failure from one your change caused.
September 28, 2026 · 5 min read
Why a sharded Playwright run can report fewer tests than it has, how to spot the gap, and how to keep results when a CI machine crashes or runs out of memory.
September 27, 2026 · 6 min read
How to detect, reproduce and fix flaky Playwright tests: retries, traces, --repeat-each, the usual causes, and tracking flakiness across runs.
September 27, 2026 · 5 min read
Playwright's built-in reporters, opening the HTML report from CI with show-report, merging sharded reports, and the questions one report can't answer.
September 27, 2026 · 5 min read
AI agents and Playwright healers write and fix tests fast. What to check in their changes, the patterns that weaken a test, and how to review them in PRs.
September 27, 2026 · 5 min read
A red Playwright run on a pull request mixes new failures, tests already failing on main and flaky tests. How to tell them apart, by hand and automatically.