Reviewing agent-generated code with an independent second model in VS Code
Review complete Git changes in VS Code, then use focused repository context when an AI reviewer needs more evidence.
Engineering Insights
Practical notes, release announcements, and engineering deep-dives from the projects built at PVR Labs.
I Taught Maven to Be an Introvert
How a small Bash wrapper reduced successful Maven output by more than 99.7% while preserving useful diagnostics for coding agents.
Air-Gapped AI Coding: Keeping Proprietary Repositories Off the Cloud
Use AI coding assistants without granting them repository access: map locally, disclose only reviewed context, and confirm every proposed write.
Lightweight Spring Boot Monitoring Without Prometheus and Grafana
Turn Spring Boot Actuator data into a focused dashboard for a few applications on a VPS, without operating a full Prometheus and Grafana stack.
Bypassing the Claude Code "Token Tax": Smart Code Reviews with Local Mapping
Autonomous coding agents consume tokens rapidly during reviews by rescanning projects and pulling broad context. Learn how to map locally, extract precise snippets, and review efficiently in any web chat.