Archive
Browse older posts.
A practical implementation of diagnosis-first self-healing end-to-end tests using Claude Code, Playwright, and GitHub Actions, with guarded repair pull requests and a complete example repository.
Read more: Self-Healing Tests with AI: Triage Before RepairA practical look at AI coding agents in 2026: critical-thinking models, agent workflows, reusable skills, costs, and the limits of open models.
Read more: AI Coding Agents in 2026: Discovery, Skills, Costs, and What Comes NextNotes from setting up Bonsai Image 4B locally on Apple Silicon, debugging the first failed path, wrapping the working pipeline in a reusable CLI, and using it from any project directory.
Read more: Local Image Generation with Bonsai Image 4BHow a local testing stack evolved from one Docker-based environment into three distinct runtime profiles with clearer intent: lightweight, full, and server.
Read more: From One Local Stack to Three Training ProfilesA practical mental model for developers: understand LLMs, AI agents, function calling, and why coding agents built on the same model behave differently.
Read more: LLMs vs AI Agents: A Practical Mental Model for Developers