Your Palette Validator Only Validates the Pairs You Run

Our CI validated the chart palette and stayed green while two lines sat at CVD delta-E 1.6 in dark mode. Validators check the pair-list you give them - and charts draw pairs you never listed.

Read more →

Thirteen CI Runs, Thirteen Real Bugs: What a Cold Machine Knows That Your Laptop Doesn't

We turned on CI for a 64-project monorepo with 10,000 green tests. It failed 13 times in a row - and every single failure was a real, pre-existing defect the fast machines had been hiding.

Read more →

The Reviewer Who Refutes: Running an Adversarial Review Cycle with AI Agents

Every AI-drafted change in our repo faces a second AI whose only job is to prove it wrong - with negative controls. The cycle has caught a real defect on nearly every pass, including bugs in my own fixes.

Read more →

The @font-face Nobody Read: Embedding Fonts in SVG-to-PNG Pipelines, Measured

My plan to make OG-card rendering portable was to embed TTFs as data-URI @font-face in the SVG. Measurement killed the plan twice — once for real, once because my probe was lying to me.

Read more →

The Day the Data Picked the Backlog — Ten Shipped Waves and One Owlbear

Search Console finally had enough data to rank my pages, so I let it write the backlog. Ten feature waves shipped in one day — and the live verification step caught a bug my unit tests never would have.

Read more →

A Practical SEO Guide, Part 5: Verify & Iterate

SEO isn't a task you finish — it's a state you maintain. How to verify every fix from this series actually shipped, and how to catch regressions before Google does.

Read more →