I tracked one real workday. I turned the repeatable parts into 12 AI skills with inputs, outputs, and safety rules. Then I open-sourced the whole kit so other operators can run the same challenge on themselves.
One workday tracked. Twelve skills built. Six templates published. Four weeks of practice. All in public, all forkable.
Vol. I · No. 01 · May 2026One worked example before the catalog. Sample input shown. Sample output shown. Eleven more specimens follow.
Turns a messy day of meetings, follow-ups, and decisions into a clean task log you can mine for patterns.
Morning leadership meeting, reviewed pipeline issues, answered three follow-up questions from the sales team, checked the Monday scorecard, researched two AI tools, drafted a LinkedIn post, sent a handoff to operations.
TASK CATEGORY WORK TYPE AUTO Morning leadership meeting Meeting Judgment-heavy Med Pipeline review Report Judgment-heavy High Sales follow-up answers Follow-up Repeatable High Monday scorecard check Report Repeatable High AI tool research Research Research-heavy Med LinkedIn draft Content Repeatable High Operations handoff Handoff Admin-heavy High
Most AI advice starts with tools.
This project starts with the job.
I wrote down how I spent one real workday. Not the job description version. The real version.
Better one-off questions aren't the move. Teaching the system how your work actually works is.
I open-sourced the kit. Twelve skills, six templates, launch copy, a community guide.
Other operators can run the same challenge on themselves.
Your turn.
Every skill below carries the same five-field spec as the featured specimen above. Job pain, tools it touches, what you give it, what it produces, and what it will not do.
Reads your workday log and surfaces the top automation candidates ranked by leverage and risk.
Turns meeting notes or a transcript into decisions, owners, deadlines, risks, and a clean handoff.
Drafts the follow-up message you've been meaning to send. Never sends automatically.
Translates business systems into plain English so any leader can understand the workflow without a demo.
Turns a report full of numbers into a one-page explanation with a headline, insights, and a recommended action.
Preps a focused, respectful coaching conversation built around one specific outcome.
Turns process notes that live in one person's head into a usable playbook with steps, examples, common mistakes, and a checklist.
Reviews public-facing content for risky claims before legal sees it. First pass only.
Tests an AI startup from a real operator's perspective and produces a useful founder message instead of a vague review.
Turns a real workday or shipped project into a LinkedIn post with hook, action, lesson, and a question.
Tries the intimidating tool you've been avoiding and produces an honest field report with what worked, what broke, and what to do next.
No theory week. No "intro to AI" week. Every week ships an artifact you can point to.
Track meetings, follow-ups, reports, decisions, research, coaching, and handoffs. Use the day-tracker worksheet. Don't pretty it up. The point is the real version of your work, not the job description version.
Each skill gets a job pain, an input, an output, a tool boundary, and a safety rule. Use the skill-readme template. Build skills you'll actually run again, not skills that sound good in a deck.
Publish the GitHub repo. Build the portfolio site with one agent card per skill. Write the LinkedIn post. The work has to be visible to count.
Search "AI for [your job]." Try every tool that comes up. Score each one. Message the founders with operator-level feedback. Try one tool that scares you. Take one tactical course. Then loop.
Six worksheets, two LinkedIn posts already drafted, and a four-week facilitator guide if you want to run this with your team or community. No email gate. One click to open.
I stopped asking AI vague questions. I started giving it my actual job.