What changed
The first issue of this newsletter was about MCP servers and stdout. That audience is still here, and so is the Playbook. What changed is who else I am writing for.
The assistants got good at the plumbing. Claude, ChatGPT, Cursor, and Codex can now search, read files, and drive the tools they need without anyone wiring them together. What they still cannot do is read your mind. The gap between a generic answer and a useful one is the part you tell them: the rules they follow every time, the skills they reuse, and requests that leave nothing to guess.
Most people were never taught any of that, and most of them do not write code. So the site now has two tracks.
Everyday is for professionals and business owners who use an AI assistant and want it to stop guessing. You need a browser and an assistant subscription. Nothing else.
Builder is for developers and operators running agentic coding tools in a terminal and a repository.
Every page says which track it serves.
Two free things to start with
10 rules that make your AI assistant stop guessing. A short primer on prompting, context, and the moment a correction should become a standing rule. Ten sections, no code, and a table of where rules live in each assistant.
The Rules Builder. Answer eight plain questions and get a rules file ready to paste into Claude, Claude Code, ChatGPT, Cursor, or Codex. No signup, and your answers stay in your browser.
Three new posts
Why your AI assistant keeps guessing. What the assistant can see when it answers, what it never sees, and a before and after that turns a four-miss email into a first-try draft.
Rules vs prompts: what to write once and what to say every time. One test for sorting instructions into the message or a standing rule, with a bookkeeper's ninety-word ritual cut to eighteen.
Your first skill in fifteen minutes. Build one reusable skill end to end, install it in your assistant, and check that it fires.
The lineup
All prices in USD. One-time purchases only: no subscriptions, no retainers. Digital products include 6 months of free updates.
Everyday
| Offer | Price | What you get |
|---|---|---|
| AI Field Guide | $39 USD | Prompting, reading what the model did, writing rules, reusable skills, red flags, and setup steps for each assistant |
| Skills & Rules Starter Kit | $149 USD | Fill-in rules files and six skill templates, each with a guide to adapting it |
| Get Set Up bundle | $549 USD | Field Guide, Starter Kit, and Guided Setup (separately $638 USD) |
| Done-For-You Essentials | $1,250 USD | Rules for one assistant, up to 3 custom skills, a 45-minute handover, a written usage sheet, and 14 days of fixes |
Builder
| Offer | Price | What you get |
|---|---|---|
| Builder Edition | $249 USD | The whole Starter Kit plus the developer chapters: install, skill design, hook safety, multi-agent dispatch, test strategy, git and CI wiring |
| Builder bundle | $649 USD | Field Guide, Builder Edition, and Guided Setup (separately $738 USD) |
| MCP Production Playbook | $29 USD | The playbook from issue one, for teams still shipping MCP servers |
| Done-For-You Complete | $2,500 USD | Layered rules, up to 8 skills, automated checks installed and verified, multi-agent conventions, a 60-minute handover, and 30 days of fixes |
Both tracks
Guided Setup, $450 USD: a two-hour 1:1 session setting up your own rules and skills, with the Starter Kit included.
The Done-For-You tiers are fixed packages defined by what you get, not by hours. Anything beyond a package is a fixed-price change order.
Where to start
If you use an assistant and have never written a rule, start with the primer and the Rules Builder. Both are free and take less time than your next bad draft.
If you already correct your assistant on the same things every week, the AI Field Guide, $39 USD is the whole method in one read.
If you want to hand the setup to someone else: book Guided Setup.
Next issue goes out when there is something worth sending.