How to Set Up Your Repo for Claude Code and Codex
Nine layers of guardrails that make AI coding agents safe, predictable, and actually useful — from CLAUDE.md to CI pipelines to specialized review agents.
Nine layers of guardrails that make AI coding agents safe, predictable, and actually useful — from CLAUDE.md to CI pipelines to specialized review agents.
The old question was 'can you code?' The new question is 'can you build and ship?' A framework for evaluating engineers who use Claude Code, Cursor, and other AI-assisted coding tools.
Test Driven Development is dead. Quality control can be largely automated now. That frees you to focus on what actually matters: building things people want.
A reusable Claude Code skill for generating SVG logos. Point it at your repo, explore multiple logo concepts, refine the strongest direction, and export production-ready icons.
Anthropic just raised $30B. A small fraction could fund the Claude Code showcase the ecosystem actually needs.

The real barriers to AI coding tool adoption aren't technical — they're management problems, employee problems, and human nature problems.

Claude's default mode is eager compliance—give it vague instructions and it'll happily make assumptions. These workflow skills force systematic iOS HIG verification, catching platform convention violations you'd never think to check.

Claude's default mode is eager compliance. These interview skills flip the script—using AskUserQuestion to force systematic questioning before any code gets written.

Four Claude Code skills that generate and execute user workflow tests using Claude-in-Chrome and iOS Simulator MCPs. Catch bugs and edge cases while you're still building.