What a command centre is

A command centre is a folder on your machine that you open in Claude Code (or Cowork). It contains markdown files that tell Claude who you are, how you work, and how to do the tasks you do repeatedly — alongside small Python scripts that handle the I/O Claude can’t do directly.

You drive it from the terminal. You type a command in plain English (“triage my inbox”, “build a deck from this outline”), Claude reads the relevant context and workflow files, executes the task, and saves outputs you review. Nothing sends, schedules, or commits without you.

This is not a new product. It’s a folder pattern. The same primitives the rest of the Playbook covers — CLAUDE.md, context files, scripts — assembled for personal operational work rather than building applications.

What the command centre is for

Three workflows recur in most knowledge-work jobs and benefit disproportionately from this pattern:

Inbox triage. Pulling unread mail, classifying by your real priority rules (not generic urgency cues), and drafting replies in your voice as .eml files you review and send manually.

Morning brief. A one-screen summary of the day’s calendar, inbox state, and follow-ups owed — produced from your context and surfaced before you start reactive work.

Document generation. Turning a markdown outline into a PowerPoint deck against your brand template, or a structured doc against a template, with the structural work done by Claude and the visual finishing done by you.

The pattern extends to anything that has the same shape: repeatable, requires personal context, produces a draft you review before acting on. Meeting prep, weekly reports, personal CRM follow-ups all fit.

What it isn’t

Not autonomous. Nothing sends, accepts, schedules, or commits to anything without your explicit review. Drafts only.

Not a UI project. The terminal is the interface. PowerPoint, Outlook, and your text editor are the output viewers. Building a web dashboard on top adds maintenance burden without value — every output already has an excellent native viewer.

Not a knowledge base. Project documents live in their proper homes (Drive, SharePoint, wherever). The command centre holds context about you and how you work — not the work itself.

Not a daemon. Don’t put this on a cron. Run it on demand, when you’re at the keyboard and ready to review outputs. Auto-running multiplies the risk of acting on stale context.

Why a folder rather than a chat or an app

A chat starts from zero every session. The command centre carries context that stays current — your priorities, the people you work with, your active projects, your voice — without you re-pasting it each time.

An app would mean a build, a deploy, a maintenance burden, and a UI that duplicates what your existing tools already do well. A folder costs nothing to build and edit; the only thing it produces are the outputs you actually want.

System, not conversation

A chat is something you have, episode by episode — every session starts from zero, and the same questions get asked again every time. A command centre is something you build and run: the questions you’d otherwise type repeatedly into chat — who is this person, what’s my voice for this audience, what’s the priority shape this month? — are answered once, in files, and applied consistently. The leap is from “open Claude, ask a question, close the tab” to “configure once, then drive it as a system.”

Design principles

Five principles shape the structure described in the rest of this section:

1. CLAUDE.md is an index, not an encyclopedia. Under 80 lines. It points to detailed files; it doesn’t contain the detail itself.
2. Context is loaded on demand. Claude only reads the files relevant to the current task. Pile everything into one file and the model starts ignoring it.
3. Workflows separate process from I/O. Markdown workflow files describe steps and rules; Python scripts handle reading and writing.
4. Drafts only, never sends. Every output goes to a file you review.
5. Real examples beat synthetic ones. Voice files, brief format, people files — all benefit from your actual content over invented content.

Compatibility

The structure works for both Claude Code (terminal) and Cowork (desktop). Both load CLAUDE.md the same way when you open a folder. You don’t need a different setup for each.

The next section walks through the folder layout. The 11 sections after that cover each file in turn — what it’s for, what to include, what to exclude, and a template.