I build tools that analyse Claude Code sessions for a living. The same patterns show up in almost every session: the same instructions pasted again, the same style guide restated, the same corrections typed by the same developers. Skills fix most of it. This course is how I show you which Skill to write first.
You might know my other projects:
The session analyser for AI-assisted development. Upload a Claude Code transcript and get scored across 8 fluency dimensions, with actionable recommendations.
A VSCode extension that monitors your Claude Code sessions in real-time: live health scores, anti-pattern alerts, and per-model cost tracking, right inside your editor.
I run a consulting practice called AIFirstPartner. I work with clients in government, legal, and enterprise to get Claude into real production workflows. I have shipped 12 production AI systems across those engagements so far. Some of them are the kind of thing you would build in a weekend and keep using for a year. Others took months and still needed three rewrites.
Alongside the consulting, I am fractional CTO at Speculo, a cyber risk software company. That role gives me a weekly view into what actually happens when you put AI features into a product with real users, real auditors, and real incident response. A lot of the guardrails I teach in this course came out of things that went wrong, quietly, in a product under load.
My day role is managing cloud and data platform engineering for a large Australian government department supporting 20+ agencies across the portfolio. Fifteen plus years of enterprise technology sit behind that. It is the reason I care about things like tool restrictions, reproducible setups, and commit-to-Git distribution for Skills. What works for one developer rarely works for a team of 200 without planning.
I publish my build process openly at 150days.aifirstpartner.com. If you want to see what I actually ship week to week before you trust me to teach you, that is the place to look. The same techniques taught in this course show up repeatedly in that log.
Every Claude Code user hits the same wall: you end up typing the same instructions into every session. "Write tests like this." "Review PRs this way." "Use these commit conventions." It's tedious, it's inconsistent, and across a team it's impossible to keep everyone aligned.
Skills solve exactly that problem — but only if you know how to write them well. A badly-described skill never triggers. A skill with the wrong tool permissions gets blocked. A skill that duplicates CLAUDE.md just adds noise.
This course shows you how to build Skills that actually activate, compose cleanly with subagents and hooks, and scale across a team through Git, plugins, and enterprise settings.
Three modules, eleven lessons, two hands-on exercises, and a capstone project where you design a complete team skills workflow:
Questions, feedback, or just want to say hi — I'd love to hear from you.
All tools built by AI First Partner