Tagged: claude-code

• 

  I Bought the Same Paprika Three Times — Then I Rebuilt Our Agent Harness the Way I Rebuilt My Kitchen

  Our repositories supported two coding agents, so every rule got written twice — once for Claude Code, once for GitHub Copilot. Two weeks ago I collapsed all of it into one .agents folder and one AGENTS.md, the convention Codex and twenty-odd other agents already read. Here is what that cost, what it bought, and why our memory folder stopped being a diary.

  Read more →
• 

  You Can Have OwlMeans Today — One Command, Any Agent

  OwlMeans Common now ships a create-app command. One line scaffolds a fully-skilled fullstack TypeScript monorepo you build end-to-end with Claude Code or GitHub Copilot — no harness to wire, no skills to write. It is a taste of OwlMeans power before the Platform is public, on the agent you already use.

  Read more →
• 

  I Stopped Building My Product for Two Months — to Build the Company Around It

  For two months I shipped no product features. I built an agentic operating system for the whole company instead — a compounding wiki, self-learning skills, every repo linked in, and a pipeline that turns an idea into a finished, illustrated post. Here is the story, why it beats running a bare agent, and how you can have the same thing on whatever AI you already use.

  Read more →