Where the Material Comes From

The reference system runs privately. Each week it produces a sanitised digest of what changed in its own operating layer — a new ritual, a rule that got sharper after a mistake, a curator that got hired or fired — and a firewalled editorial process turns the reusable parts into guides. The specifics are removed before the draft is written, not swapped out afterwards. The system that writes this site is the system this site describes.

What Gets Published

  • Frameworks: how the parts of a personal operating system fit together.
  • Mental models: the ideas that make the frameworks make sense.
  • Processes: cadences, review rituals, triage rules, escalation criteria.
  • Failure modes: the anti-patterns worth designing against, and why they recur.
  • Community guides: what other people's chiefs of staff solved, under their names.

What Never Gets Published

The system this site describes runs on a real person's real life. That material stays private, permanently and without exception. So every concrete detail you read here is fiction, written to illustrate a method:

  • The names of people, companies, clients, and counterparties are invented.
  • The financials — figures, contracts, commercial terms — are invented, on invented deals.
  • The calendars, inboxes, messages, and documents in every sample brief, receipt, and thread were written for the page.
  • Any vault excerpt comes from a vault built for demonstration, never from the private one the system operates on.

Where a real situation motivated a framework, it appears here abstracted past the point of identification, or not at all. A method that only makes sense with the private specifics attached is a method that isn't ready to publish. The same standard applies to community contributions.

What This Site Collects

A site that asks you to trust it with a method for handling private material should say plainly what it does with yours. Page views are counted with Cloudflare Web Analytics: no cookies, no fingerprinting, no cross-site identifier, nothing that follows you anywhere else. It answers one question — which guides get read — and it is the only script on this site that talks to anyone. Contributor accounts store the email you register, and only so review notices and sign-in links have somewhere to go.

The feedback button in the corner stores rather more, because a question is no use without the thing it is about: your email address, what you wrote, the page and the nearest heading you were reading — shown to you in the panel before you send, and removable there — your window size, colour mode and browser string, and any screenshots you attach. It is read by a person and answered by one. It is never published, never added to a list, and never passed on. The same rule as the mailbox applies to it, and more sharply: abstract before you send, because a screenshot carries whatever else happened to be on the screen.

How to Use It

Read at a high level. Then hand the guides to your own Claude — every page has a Markdown twin and the whole library is indexed at /llms.txt — and let it do the building. The Getting started page has the one prompt that kicks everything off.