PS C:\DevProjects\DevBlog> Get-Content .\Thoughts\AgenticMidYear2026.md Halfway through 2026, the conversations I have about agentic development have started failing in a particular way. Two developers use the same words (agents, tooling, cost, quality) and mean entirely different things by them. Neither is wrong. They are reporting from different places on a curve...
I couldn’t find a VS Code dark theme that fit two requirements I cared about: comfortable for long coding sessions, and readable for colorblind developers. The marketplace options I tried either dimmed comments past the point of usefulness, relied on red and green to carry meaning, or pushed contrast hard...
TL;DR Software fundamentals matter more now than they ever have, agents made skipping them more expensive, not less. Coding was the bottleneck. Agents made coding fast, for prototypes. Past prototype scale, the slower work (design, naming, structure) is the precondition for the speed compounding instead of collapsing. Six quotes from...
If you’ve ever had an AI code review that missed something obvious, or gotten a plan that seemed solid until implementation revealed hidden assumptions, you’ve encountered the blind spot problem. Every model—even the best ones—has blind spots. The real question isn’t which model is perfect. It’s how do we mitigate...
PS C:\DevProjects\DevBlog> Get-Content .\Architecture\StopHookQualityGates.md 🎯 Introduction In a previous post, I wrote about feedback loops through compilation, static analysis, and testing — mechanisms that let AI coding agents validate their own work iteratively. Those loops operate at the code level: does it compile, does it pass analysis, do the tests...