! 6 posts tagged
The more autonomy you give an AI agent, the more alignment you need upfront. Without shared understanding, the agent fills every gap with training-data defaults — and builds the wrong thing confidently.
Switching Claude Code to Opus 4.7 on Bedrock returned a permission error on the US and global inference profiles but worked on the EU one. Here's the reproduction, the fix, and what to tell AWS support.
Vibe coding is great — until your project outgrows a single conversation. For complex, long-lived systems, spec-driven development with OpenSpec gives AI assistants deterministic input instead of fuzzy chat history.
The architecture of an AI agent maps onto a traditional computer. The LLM is the CPU, the agent runtime is the OS, and skills, commands, and MCPs are the applications you install.
How a .ai/ directory and project constitution turned ad-hoc AI coding sessions into a repeatable engineering workflow.
Migrated the blog from Ghost to a custom Astro site built entirely with Claude Code. Went with a Cisco CLI-inspired dark-mode design — syslog timestamps, IOS command prompts, terminal aesthetics. The goal was readability first while keeping the network engineering feel baked into the UI.
The whole process was surprisingly straightforward. Astro’s content collections handle MDX well, and having an LLM pair on layout and component decisions made it fast to iterate. Happy with where it landed.