Zachary Gray

Rocky's morning brief and to-do board, rebuilt so anyone can run them.

Try the demo
git clone https://github.com/zacharytgray/daybook.git
cd daybook
python3 bin/daybook demo --fresh

Python 3.11 or newer. No accounts, no network.

  • Public, MITStatus
  • 2026Started
  • Agents and toolsKind
Code on GitHub

What it is

The code is public, and you can run it yourself.

Every morning Rocky texts me a brief of my day, and a board on my Mac mini shows what it is working on. Daybook is the open version of both, for anyone who works with AI agents.

The brief turns a calendar, a task list, a class folder and the agent's own session notes into one page. The board stays live. I mark things done there so tomorrow's brief skips them, and I hand any to-do to an agent with suggested settings.

Anything that could act on the world is off until you turn it on, and the demo runs on an invented day with no accounts. It's plain Python with no dependencies.

Two pages

One you read in the morning, and one you keep open after.

The brief

One page each morning. Code gathers your calendar, mail, tasks, classes, pull requests, agent session notes, weather and AI news.

A model writes only the words. Code draws the rest: a print of the day made from your calendar and forecast, a timeline with open blocks and overlaps, ranked next actions, and five small puzzles.

The board

It shows today, what your agents are doing, and the machines they run on. Mark a to-do done and tomorrow's brief skips it.

Hand a to-do to an agent and it comes with suggested settings, like which runner, which folder and how far it may go, plus a box for a note.

Try the demo

Python 3.11 or newer. Nothing to install, no accounts, no network.

Run it
git clone https://github.com/zacharytgray/daybook.git
cd daybook
python3 bin/daybook demo --fresh

The board and the brief both open in your browser, on your own machine. The README has the addresses.

The demo is a fixed Tuesday for an invented grad student named Avery, whose agent is called Pip. Its clock is frozen at 10:20 AM, so it looks the same every time. Marks and hand-offs work, but nothing leaves your machine. With no transport set up, Start gives you the request text to copy.

Your own day

Six steps from the demo to your real calendar.

  1. Settings

    Copy .env.example to .env and set at least your time zone, your name and your agent's name. Everything else is optional and off until set.

  2. About you

    Write prompt/about.md, a few lines on who you are and what matters in your mail and calendar. It stays out of git.

  3. Sources

    Point the sources you have at their folders, like school, projects and session notes. The customization doc walks through each one.

  4. A brief

    python3 bin/daybook brief generate runs Claude Code once, read-only, with your calendar, mail and Todoist connectors.

  5. The board

    python3 bin/daybook board serve, then open the board in your browser.

  6. Every morning

    Schedule the brief with cron, launchd, systemd or your agent's own scheduler.

What it won't do

Nothing happens on its own that you didn't switch on.

Off until you say so

  • It sends nothing. Messages, webhooks, Todoist writes and stopping or reviving a session each need a setting, and the demo forces all of them off.
  • It doesn't listen beyond your machine. Both servers stay on 127.0.0.1 unless you change that.

Even when it's on

  • The brief's model run can't write anywhere. It gets a read-only tool list and a denylist of every write tool, and it won't bill an API key unless you allow it.
  • A hand-off is a request to your agent, not a command. Your agent decides what to run, under its own rules.

Go run it

The repo has everything on this page, and the rest.

Sit at the desk, in 3D(the 3D version)There's a 3D version of this site. It needs a bigger screen than this one.This browser can't run WebGL, so it stays flat.It moves a lot, and your system asked for less motion.
NextPublic template

UTulsa Course Brain

Agents connected to my academic life, with my lectures, Blackboard, and deadlines in one repo my AI can read.