I Built a Test Bench for Website Design Skills
Thirty-two generated websites, six models, three design systems, and a grading layer built to separate skill fidelity from attractive screenshots.
Build log documenting projects, experiments, lessons learned, and releases — from shipped Chrome extensions to local AI orchestration systems and developer safety tools.
Detailed logs, architectural notes, and feature progress from active projects.
Thirty-two generated websites, six models, three design systems, and a grading layer built to separate skill fidelity from attractive screenshots.
Localhost Watchdog moves from inspection into guarded management with evidence-based stop, start, restart, adoption, and a Windows tray companion.
A Chrome extension and local companion that turns PageSpeed Insights into structured Markdown with confidence thresholds, fixed recommendation ordering, multi-run median testing, and variance analysis — ready for coding agents to act on.
An experiment in local AI orchestration — a Local Brain, Task Tracks, AI Pit Crew, capability routing, and optional NearbyNodes working together so one person can operate like a coordinated team using small local models.
Windows-first port scanner and process manager with dry runs, confidence gates, PID revalidation, and no destructive action without confirmation.