The Series
Field notes from a daily practice with AI coding agents. Patterns, tools, sharp edges. Evidence-based, no fluff.
The Reviewer Told to Write Nothing: Independence as a Prompt Constraint
'You are an INDEPENDENT REVIEWER. You did not write this code. Change nothing — no edits, no commits.' Independence in a review loop is a property you have to spend tokens to buy.
The Keyboard Trap Video.js Hid: A Control Bar You Cannot Tab To
WCAG 2.1.1 and 2.4.3 both fail when a player hides its own control bar from the keyboard. The fix was three lines of CSS and one line of reasoning.
Read Exists But Is Not Enabled: The Difference Between Absent and Withheld
A tool that does not exist and a tool that exists but is switched off produce different error text, and only one of them is a configuration problem you can fix.
Ranking by the Only Field You Have: When the Schema Silently Picks Your Metric
Two independent agents, same model, same session tree, reached the same conclusion within minutes: the fields they needed to rank by did not exist, so they ranked by message count.
What we ship
ValidationForge proves that code works. Anneal proves the plan works before any code gets written. Two flagships, both production-grade. The other cards are companion tooling from the series.
Get the next field note before it ships.
4,206 engineers, operators, and weirdos. No spam, ever.