Blog · 15 runs logged

The crash log

Field notes from breaking software on purpose: Playwright patterns, CI pipelines, and the testing strategy behind them.

Map of the crash log

15 posts · 16 internal links · grouped by first tag · dot size = reading time

Each dot is a post; lines are links between posts. Bigger dots take longer to read.

Playwright
Testing Strategy
AI
BDD
- Playwright BDD: Combining Cucumber with Page Objects 1 link
GitHub Actions
- How to Run Preview Environments per PR With a Fixed Pool of Slots 2 links
Mocking
- The Golden Rule of Mocking: Mock the Boundary, Not the Code 2 links
Nix
- Nix Flakes for SDET Setup 2 links

15 of 15 posts

Run 015 Oct 10, 2026 6 min read

The Golden Rule of Mocking: Mock the Boundary, Not the Code

Mock the payment provider at the HTTP boundary, not your own client class. A TypeScript and MSW example of tests that can actually catch a Stripe bug.
- Mocking
- Testing Strategy
- Test Architecture
- TypeScript
Run 014 Oct 10, 2026 7 min read

Test Database Isolation in pytest: Reads Share, Writes Get Their Own

Share one test database and tests leak state into each other. Give each its own and the suite crawls. Match pytest fixture scope to what each test does.
- Testing Strategy
- Test Architecture
- Pytest
Run 013 Oct 10, 2026 7 min read

How to Sandbox Your AI Coding Agent with a Dev Container

Run your AI coding agent without permission prompts but inside a dev container, with deny rules, read-only mounts, an egress allowlist and no prod credentials.
- AI
- Developer Environment
- Testing
Run 012 Oct 10, 2026 8 min read

How to Cut QA Busywork with Rules Files, Agent Skills and MCP Servers

A rules file for house style, skills for routine workflows, an MCP server for Jira: copy-paste setup for Claude Code and Cursor, with writes on a leash.
- AI
- QA
- Playwright
Run 011 Oct 10, 2026 9 min read

How to Run Preview Environments per PR With a Fixed Pool of Slots

A live preview URL for every pull request without a backend per branch: a fixed slot pool, a GitHub Actions claim step, Playwright on the slot, auto-release.
- GitHub Actions
- CI
- Playwright
Run 010 Oct 10, 2026 5 min read

Coverage Is a Health Signal, Not a Per-PR Gate

Making every PR raise the coverage number rewards padded tests. Gate the patch, read the project trend, and coverage starts telling you the truth.
- Testing Strategy
- CI
- GitHub Actions
- Code Coverage
Run 009 Oct 10, 2026 8 min read

Every Integration Test Has the Same Five Parts

Two questions to ask before you write a test, then the five parts every one shares, worked through a create-payment test that checks the ledger.
- Testing Strategy
- Playwright
- GraphQL
- TypeScript
Run 008 Oct 9, 2026 8 min read

Playwright Page Factory + Composition: Scaling a Vanilla POM

Combine a page factory, composition pattern, skipGoto navigation, and a Playwright fixture, so specs stay thin as the suite grows.
- Playwright
- Page Object Model
- TypeScript
- Test Architecture
Run 007 Apr 23, 2026 4 min read

AI Code Smells: A Review Checklist

AI coding assistants fill gaps with plausible-looking code. Nine recurring AI code smells to catch in review, and what to look for when you spot each one.
- AI
- Code Review
- Testing
Run 006 Mar 14, 2026 6 min read

How to Test Feature Flags Without Testing 2ⁿ Combinations

Every feature flag doubles the test surface: n flags means 2ⁿ configurations. A QA view of why flags are hell to test, and how pairwise testing helps.
- Testing Strategy
- Feature Flags
- QA
- Pairwise Testing
Run 005 Nov 2, 2025 8 min read

Vanilla Playwright with Page Object Model: A Clean Setup

A Playwright Page Object Model layout with no BDD and no BasePage: one pageFactory owns navigation, page objects own interactions, specs read like scripts.
- Playwright
- Page Object Model
- TypeScript
- Test Architecture
Run 004 Oct 26, 2025 6 min read

Run Playwright Tests from a Separate Repository in CI

Keep Playwright tests in their own repository and run them on every PR in your app repo with GitHub Actions, a dual checkout and a least-privilege token.
- Playwright
- GitHub Actions
- CI
Run 003 Oct 17, 2025 4 min read

Playwright BDD: Combining Cucumber with Page Objects

Combine Cucumber/Gherkin with the Page Object Model in Playwright so BDD suites stay maintainable instead of decaying into scattered utility functions.
- BDD
- Playwright
- Page Object Model
- Cucumber
Run 002 Oct 16, 2025 4 min read

Playwright: Interacting with Closed Shadow DOM in iFrames

Force closed shadow roots open with Playwright addInitScript to test web components inside iframes, like embedded payment or ID-verification widgets.
- Playwright
- iFrames
- Shadow DOM
- Web Components
Run 001 Oct 13, 2025 2 min read

Nix Flakes for SDET Setup

Set up a reproducible SDET workstation on macOS, Linux or WSL2 in about five minutes with Nix Flakes, then reuse it in CI with nix develop.
- Nix
- Developer Environment
- CI

Map of the crash log

The Golden Rule of Mocking: Mock the Boundary, Not the Code

Test Database Isolation in pytest: Reads Share, Writes Get Their Own

How to Sandbox Your AI Coding Agent with a Dev Container

How to Cut QA Busywork with Rules Files, Agent Skills and MCP Servers

How to Run Preview Environments per PR With a Fixed Pool of Slots

Coverage Is a Health Signal, Not a Per-PR Gate

Every Integration Test Has the Same Five Parts

Playwright Page Factory + Composition: Scaling a Vanilla POM

AI Code Smells: A Review Checklist

How to Test Feature Flags Without Testing 2ⁿ Combinations

Vanilla Playwright with Page Object Model: A Clean Setup

Run Playwright Tests from a Separate Repository in CI

Playwright BDD: Combining Cucumber with Page Objects

Playwright: Interacting with Closed Shadow DOM in iFrames

Nix Flakes for SDET Setup