AI & Tooling
Building with AI coding tools: Claude Code plugins, skills, agents, and the workflows that grow up around them.
Running Codex inside Claude Code
Why you'd want OpenAI's Codex reviewing Claude's work from inside Claude Code, the plugin that bridges them, the review-and-rescue workflow, and two copy-paste skills that make delegation a habit.
llms.txt, and who actually reads it
A plain-text file at /llms.txt hands language models a curated map of your site. Here is how to add one, what it is genuinely good for (coding agents), and why you should not expect it to move your AI-search numbers.
The grill-me Skill for Claude Code
A look at Matt Pocock's tiny, viral grill-me skill: instead of coding your idea straight away, Claude interviews you one question at a time until the plan has no ambiguity left.
Loop Engineering in Claude Code: Designing the System That Prompts the Agent
Loop engineering is the shift from prompting an agent step by step to designing a system that prompts it for you. Here is what that means in Claude Code, the building blocks it already ships, a worked example, and an honest look at when it is worth the tokens on a subscription.
Claude Code Hooks, by Example
CLAUDE.md asks Claude Code nicely. Hooks make things happen for certain. A short, practical tour of how hooks are wired, built around two that I actually run: one that installs dependencies on session start, and one that filters noisy command output before it reaches the model.
Getting More Out of Claude Code: Prompting and Token Economy
A practical guide to driving Claude Code well: how to prompt for fewer wrong turns, how to keep the context window healthy, and a straight answer to whether running Opus 4.8 at max effort on everything is actually a good idea.
How to Build a Claude Code Plugin
A hands on guide to packaging your Claude Code workflow as a plugin. We build a real conventional-commits helper with a slash command and an auto-invoked skill, then publish it through a marketplace and version it for release.