Articles on
reliable software.
I share what I learn about software quality, AI-assisted coding, and working with agents.
Browse articles
All articles
I Made My App Agent-Friendly with WebMCP
How I added WebMCP tools to my social media planner so AI agents can find drafts, create posts and update schedules through an open browser tab.
Creating rules that stick
Why does your AI agent keep ignoring your rules? Here's what I've learned about writing better instructions and checking whether they're actually followed.
How to Make Your SKILL.md Interactive
How I added interactive questions to a skill so it works in Claude Code, Cursor, and Codex, and what still breaks across harnesses.
Maslow's Hammer and Three Lies QA Tells Itself
A short post on what needs to change for QA to thrive in the age of AI.
AGENTS.md Files Are Not Misleading Your AI Agent
A recent study sparked headlines claiming AGENTS.md files hurt AI agent performance. But the full picture tells a different story - human-written context files actually improve results by 4%.
Don’t let A.I. read your .env files
Stop storing API keys and database passwords in plain text .env files. Learn how to use 1Password CLI to inject secrets at runtime for better security.
How I automated the annoying part of my job with Goose and Playwright MCP
Learn how to automate repetitive issue creation tasks using Goose desktop app and Playwright MCP. Stop context switching and let AI handle the boring stuff.
10 Tips for Writing Playwright Tests with Cursor
Practical tips gathered from months of writing Playwright tests with Cursor. Learn how to use project rules, workflows, screenshots, and MCP for better test automation.
2FA testing with Playwright and Mailosaur
Learn how to automate email magic links, SMS verification codes, and authenticator app logins in Playwright using Mailosaur for comprehensive authentication testing.
Testing Will Become More Important, Not Less
LLMs speed up coding but quality concerns remain. Testing becomes more crucial, evolving to include AI-generated tests while human expertise in test design stays valuable.
10 of 70 articles