-
Agent Skills: The Four Bytes Four Gates Missed
Four gates read the same twenty-two agent skills and none caught four bad bytes. Why coverage comes from changing the mode of inspection, not adding gates.
-
What the JetEngine REST API actually reaches
Our note said the JetEngine REST API was gated behind approval. A second route answered 200. What it reaches, its limits, and why the note went…
-
The Slide Deck Check That Read Four-Pixel Text
A slide deck check that reads declared font sizes passes text four pixels high. Why legibility and traceability must be measured on the surface readers see.
-
Design Tokens Are a Count, Not a Convention
A design token file does not make a system. Why adherence has to be counted off the rendered page, in both themes, and why the only…
-
Our content audit measured the wrong surface
A content audit of our own site: three articles failed the same gate after three reviews, and the checker was measuring stored text, not the rendered…
-
Why subscription reconciliation closes on a stale number
A subscription reconciliation that never subtracts anything is not one. Why a cycle close must difference twice, and the repair order that survives contact.
-
Twelve Parallel Agents Reported Success on Wrong Data
Twelve parallel agents each reported success on a harvest that was quietly wrong. What a slice-level report cannot see, and the reconciliation that can.
-
When a payment webhook lies about the money
A payment webhook acknowledgement proves delivery, not durable state. The four surfaces that return success while the money state they imply does not exist.
-
Stale but true: the hardest bug in AI agent memory
For weeks our own canon carried a note saying that a particular endpoint was a useless dead end. An agent had tried it, failed, and written…
-
AI Code Review Gave Five Stars to a Truncated File
A judge model returned 5-5-5-5-5 on a file that stopped in the middle of a section. The final heading had nothing under it and the last…