Posts

• What Happens When You Press Enter

  The agent loop explained from first principles. What the model actually does, how context is assembled before you type, and why context engineering — not prompt engineering — is the skill that matters.

  April 5, 2026 · 11 min read

• Instructions and Rule System

  Anthropic calls CLAUDE.md files 'instructions,' not 'rules.' This post explains the distinction, how both mechanisms work, and why the ecosystem's naming confusion matters.

  March 30, 2026 · 8 min read

• My Projects With ESLint Stay Clean. The Rest Drift.

  In AI-assisted development, ESLint's architectural boundary enforcement is the difference between a codebase that stays clean and one that silently drifts. Here's why specs alone aren't enough.

  March 24, 2026 · 5 min read

• I Built This Blog by Prompting Claude Code

  I built this blog entirely by prompting Claude Code — no file editing by hand, no manual git workflow. This is the story of how, and the stack that made it work: a Clojure SSG, vault-sourced content, devtools-mcp, and superpowers.

  March 21, 2026 · 5 min read

• Building Your AI Toolkit

  An agent is a loop around a language model. The five building blocks — CLI, MCP, Skills, Rules, and Hooks — and why the role shifts from writing code to designing workflows.

  March 17, 2026 · 9 min read

• Hello World

  First post — what this blog is about, who's behind it, and the Obsidian + Clojure stack that powers it.

  March 16, 2026 · 2 min read