Free
The Claude stack, in one reference.
Seven sections, no filler, no email wall. A genuine slice of the course material, published free because the people who read it and want more are exactly the right people for the cohort.
Contents
Seven sections.
Written for someone who already uses Claude and suspects there is more to it.
The surfaces, and which to use when
Six ways to reach the same models. A short decision rule for each, so you stop building on the wrong one.
The instruction file
What belongs in it, what quietly makes results worse, and why most of them are three times longer than they should be.
Context and memory
How the window actually gets spent, what persists between sessions, and the failure mode where more context makes output worse.
Skills and MCP
Teaching Claude your tools instead of re-describing them every session. When a skill beats a longer prompt, and when it is overkill.
Agents that run unattended
The difference between a chat that helps and a process that runs. Guardrails, because unattended means unwatched.
What it costs
Honest token math. Where spend actually goes, and the two habits responsible for most surprise bills.
A starting checklist
The setup to do first, in order, in about an hour.
Why free
Because reading it is not the hard part.
Everything in the toolkit is accurate and useful, and it will still leave you with the gap this whole site is about: knowing what is possible is not the same as having it running against your own work. If the toolkit is enough on its own, genuinely take it and go. If it makes the gap more obvious, that is what the cohort is for.
Reading it is one thing. Building it is another.
The cohort is the same material, applied to your work, with someone watching your screen when it breaks.
Next cohort: Sep 15 – Sep 16, 2026 · 4 of 20 seats left