Tell your AI once.
A workspace of plain files your assistant reads before it answers. It remembers what you decided, keeps each subject separate, and tells you when something it knows has gone out of date.
Works withClaude CodeCodexOpenCodeand any tool that reads AGENTS.md
- AGENTS.md
- PROFILE.md
- COMPASS.md
- work/
- product/
- hiring/
- studies/
- .agents/
You: Plan my week. Your assistant reads 3 files, then answers: The release is on the 14th, so Monday and Tuesday go to the checklist. Thursday's call needs the pricing deck, still in review. Two interviews fit Wednesday afternoon.
How it works
Your context, in files it can trust.
Ask it to plan your week. It opens the files the task needs, and only those, then answers without asking you anything.
Every file carries the date it was last checked. Past it, the assistant asks before it plans around something stale.
Say hello. It interviews you for fifteen minutes and builds your first areas from what you say.
When the kit gets better, the changes come in one file at a time, with your yes. What you wrote is never touched.
Plan my week.
The release is on the 14th, so Monday and Tuesday go to the checklist. Thursday needs the pricing deck.
Plan the podcast launch.
This file was last checked on 3 March. Before I plan around 15 April: is that still the date?
My studies and a side project.
Got it. I have set up an area for each. Now a few questions about you.
Upgrade the workspace.
Two clean updates, and one file you also edited. Want to see that one both ways?
The rules
Six rules keep it true.
They came from daily use since March 2026, and a few from things going wrong.
- 1
One folder per area
Each area of your work or life gets a CONTEXT.md that says what is true now.
- 2
Every file says how old it is
Past its review date, a file is suspect, and the assistant says so instead of trusting it. A small linter flags what is overdue, broken links and copied dates.
- 3
One fact, one home
A deadline lives in exactly one file. Everything else links to it.
Learned when a summary kept a client deadline for eleven days after it changed.
- 4
Decisions are logged
An append-only DECISIONS.md per area. Settled questions stay settled.
Learned when an agent filled gaps with decisions nobody had made.
- 5
A weekly rhythm you choose
GTD, Agile Results, 12 Week Year, the Compass or others, picked at onboarding.
- 6
Tasks live where you keep them
Your task app if your assistant can reach it, or a TASKS.md. These files hold the why.
The kit
Everything is a file you can read.
AGENTS.md
The rules every agent follows, whatever the tool: read context first, check freshness, never copy a fact, update what you invalidate. Onboarding adds how you want it to work and write for you, and what it may do on its own.
Start
Fifteen minutes, by talking.
- 1Copy the templateUse this template on GitHub and keep your copy private, or download the zip.
- 2Say helloOpen the folder in your AI tool. Onboarding starts by itself.
- 3Review it weeklyYour assistant runs the upkeep, then the rhythm you chose.
An onboarding conversation:
You: Hello
Your assistant: Hi. What do you want this workspace to help with?
You: My studies and a side project.
Your assistant: Got it. I'll set up an area for each, then ask a few questions about you.
Created PROFILE.md, studies/, side-project/.
Questions
Before you start
Doesn't my AI already have memory?
Some tools remember things between chats, but that memory is hidden, stays inside one app, and tends to pull in things that have nothing to do with what you are working on. A workspace is plain files you own: you can read every word your assistant knows, fix what is wrong, keep each subject separate, and take it all to any tool.
Do I need to code?
No. It is a folder of Markdown files and a few small scripts, and your AI assistant does the setup with you. The scripts run with uv, a free tool that runs Python scripts. Nothing else to install.
Which AI tools does it work with?
Claude Code, Codex and OpenCode, and any tool that reads AGENTS.md, the shared file most agents already look for. The kit's README shows what each one does with the folder.
Do I need Notion?
No. If your task app has a way for your assistant to connect, it uses that; if not, a TASKS.md file in the folder holds your tasks. You choose at onboarding. Either way, the dates you act on live there and nowhere else.
How long does setup take?
About fifteen minutes. Open the folder in your AI tool and say hello: it interviews you, creates your first areas from your answers, connects to where your notes and tasks already live, helps you pick a weekly rhythm, and ends by doing one real task for you. If you stop halfway, it picks up where you left off.
What happens when the kit improves?
Say “upgrade the workspace”. Your assistant shows what changed since your copy was made and brings it in one file at a time, with your yes for each. A file you edited that the kit also changed is shown to you both ways, and you choose. Your profile, your areas and your answers are never touched, and the weekly review tells you when there is something new.
Is my data sent anywhere?
The kit is plain files on your own computer. Whether anything leaves it depends only on the AI tool you choose to open them with.
Want help setting it up?
Write to me. I'd also like to hear what you changed.
Hear about the next one
An email when I release something new. No schedule.