Set up your repos
Once your mind is more than a pasted file, it lives in a repo — and so does the work it does for you. You’ll usually have two kinds, and the good news is they all behave the same way by default. You set rules on a repo only when you want that one to behave differently.
Two kinds of repo
Section titled “Two kinds of repo”- Your mind — your fork: your voice, judgment, and knowledge, plus the surfaces it runs on (which tools, and what each one can do). One portable repo that is you. (Pick how deep it installs in Quickstart; put it on tools in Put your mind on your tools.)
- Your projects — the actual work: an app, a game, a site. Ordinary code repos your mind operates on.
Your mind can sit beside your projects (its own fork, working on them — the usual
way) or be mounted inside one (a .mind folder in the project — the submodule
option in Quickstart, for when your mind lives inside a bigger
project). Either way, the picture below is the same.
The easy button: one default workflow, every repo
Section titled “The easy button: one default workflow, every repo”You don’t run git. In any repo, you speak in outcomes and the agent does the plumbing:
“Draft this change.” → it works on a copy. “Publish it.” → it makes it real, and confirms it’s actually live before it says done.
Out of the box every repo follows the same shape — the one from How work flows:
- One home of record per repo. Each repo has a single place the truth lives; the agent always lands your work there, never strands it on a side copy.
- Drafting → Up for review → Live. You always see where a job is, and Live only shows once it’s truly live.
- It stops at exactly one door — right before anything becomes real (published, deployed, changed for keeps). Nothing irreversible without your yes.
That default is deliberately boring: it works the same way in your mind repo and in every project. No per-repo setup needed to get going.
The glass sliders: what you change per repo
Section titled “The glass sliders: what you change per repo”Same workflow, different settings — and every setting is a knob in plain words, never a config wall (Simple by default):
| The knob | Slide from… | …to | Set it per repo like |
|---|---|---|---|
| The yes gate | ”show me, ask first" | "just run it" | "On my game project, always ask before anything goes live.” · “On scratch notes, don’t bother me.” |
| Install depth | a pasted file | fork → compiled fleet | Quickstart — climb a rung when you want it to stick |
| Surfaces | one tool | every tool you use | Put your mind on your tools — add a tool in one line |
You set a knob once, per repo, in plain words — you never rewrite the workflow. A repo you say nothing about just uses the good default.
What’s default — at a glance
Section titled “What’s default — at a glance”| Repo | Out of the box | The one stop |
|---|---|---|
| Your mind | remembers as you go; drafts a sharper you between sessions | a change to who you are waits for your yes (how your mind gets smarter) |
| A project | the agent drafts, tests, and lands your change to that repo’s home of record | publishing / deploying waits for your yes |
You’ll have a few repos — your mind and the work it does. Every one runs the same simple workflow by default, shows you where each job is, and stops before anything goes live. Want one to behave differently? Slide a knob in plain words. You never set up a workflow — you adjust one that already works.
- How work flows here — the default workflow in full, and the yes gate
- Repo layout — every folder inside one repo, when you want to look