Report · this morning Ready.

5 of 5today

  1. Newest Claude Code is 2.1.296, shipped October 9, 2026. This machine has it.
  2. Your rules file: opened. The first line matches. 184 lines, under Anthropic's target of 200.
  3. Your memory index: opened. 143 bullets, the same as yesterday.
  4. A session can start five ways: startup, resume, clear, compact, fork.
  5. Anthropic lists 33 hook events today. Every hook in your settings is one of them.

NoteYour memory index is 190 lines. Anthropic reads only the first 200. Trim it before it gets cut.

NoteOne setting name in your settings file is not on Anthropic's list. Unverified whether it still does anything.

Anthropic changed the hooks page yesterday; nothing here disagrees.

A sample morning. The Anthropic numbers are live when the app can reach them; the machine numbers are the sample's.

For Claude Code · every morning

Before Claude works for you, it takes a short quiz.

The questions are about your own rules, your own notes, and what Anthropic's pages say today. The answer key comes from Anthropic's pages, so when they change something, the quiz changes the same morning.

We wrote it down for a reason. But if you don't look, that's the bigger issue.

curl -fsSL https://startsuperintelligence.com/install.sh | sh

One paste installs it. Then type /quickstart in Claude Code. Two files, one folder, nothing else. What it installs

Who it is for

Anyone who wrote something down for Claude and wants it read.

  • 01
    You have a rules file and a memory folder. It has to find the right instructions.
  • 02
    You do not have time to follow every change Anthropic ships. But for the rest of us that don't have time to do that, how do we know things change?
  • 03
    You are not at the keyboard all day. Even more important for the person who is not a daily power user.

It'll fit to whoever's computer it is.

The questions are the same for everyone. The proctor finds your files on your machine, and the answers come from them. Nothing here depends on how your folders are arranged.

How it works

Four steps. The last one runs by itself.

  1. 01

    Install once

    One paste puts a small skill in your Claude folder: a skill card and a short script called the proctor.

  2. 02

    Every morning, Claude checks in

    Type /quickstart. The proctor reads your rules file, your memory index, your settings and your skills, and sends only counts and fingerprints.

    You wake someone up, they see what time it is.

  3. 03

    Claude opens the files and answers

    Five to eight questions in plain words. A blank is wrong. "Could not open" is wrong. Right answers mean the file was really read.

    We're not trying to trick it.

  4. 04

    The answer key keeps itself current

    Each morning the app reads Anthropic's own Claude Code pages. When a page changes, the key changes, and the report says what moved.

    Set it and forget it.

That way, by taking this quiz, we can never have any more backtracking.

The report is short: the score, anything in your setup that disagrees with Anthropic's pages, and what changed since yesterday. We use it on our own machines every morning. We're just going to use it first.

What Anthropic changed

The last three days, straight from Anthropic's pages.

The app reads eight Claude Code pages every morning and keeps the lines that matter: hooks, memory, skills, settings, scheduled tasks, versions.

  • Looking at the last three days…

Privacy promise

The app never receives file contents — only counts, dates, version strings, setting names and fingerprints.

The proctor is short enough to read. It never sends a file's words, a file's name, an email address or a key. There is no login and no email to take the quiz.