Skip to content

init

skill

twenty seconds of setup, then out of your way

Sets MasterMind up for one project by reading your stack instead of interrogating you, and asks at most a single question.

how it fires

You never type a command. Describe your problem in your own words and MasterMind recognizes it and applies this skill. Power users can invoke it directly with /init.

the rule it routes on
Use on the first substantive work in a project MasterMind isn't set up for yet, when the user says "init", "initialize", "set up MasterMind", "onboard", "get me ready", or when no field pack matches the project's detected stack.

🧠 MasterMind ▸ init

what this skill does

This is the real text MasterMind follows: generated from the source, not a summary of it.

The problem this solves

Every tool that wants to be useful has a setup problem. To help well it needs to know things about your project: but the moment it starts asking, it becomes the obstacle it was meant to remove.

The usual result is a wizard: eight questions, half of them technical, most with answers already sitting in your project files. People abandon setup, or click through it randomly, and the tool ends up configured wrong anyway.

Init is deliberately the opposite: detect everything readable, ask at most one thing, then start working.

It runs once per project. Once MasterMind knows your stack, it never does this again.

What goes wrong without it

  • Generic advice. Without knowing your framework, versions, and conventions, an AI defaults to the most common answer on the internet, which is often two versions out of date and shaped for a different project than yours.
  • The wrong conventions imposed. It writes code in its preferred style rather than the style already in your repo, and every review turns into re-explaining your own house rules.
  • Being asked things it could have read. “What test runner do you use?” is a question with an answer sitting in a config file. Asking it wastes your time and signals the tool isn’t paying attention.
  • A polished front end on an unguarded back end. Half-configured setups tend to load knowledge for the visible layer only, leaving auth, data, and permissions on guesswork.

How it actually works

There are two paths, and MasterMind picks based on whether your folder has code in it.

If the project already exists, it asks you nothing. It reads your package files, lockfiles, configs, folder shape, database, and test setup, then loads the matching body of domain knowledge. If the closest match is imperfect, say the stored knowledge assumes React and you’re on Svelte, it prunes what doesn’t apply and fills the gaps rather than forcing a bad fit. If there’s no knowledge for your field at all, it offers a one-time deeper setup and tells you plainly what it costs: a few minutes now, reused forever after.

If the folder is empty, there’s nothing to read, so it asks one open question: what do you want to build? You answer in your own words. Technical or not, both are fine. If you name a framework, it uses it. If you only describe the product, it chooses the stack itself, states the choice with a one-line reason, and proceeds. You are never handed a menu of technical options to pick from.

It never blocks your first result. If you’re eager to start, MasterMind begins the actual work and runs any slower setup in the background. Setup is not allowed to become the thing you’re doing.

Both paths end with a short summary: detected stack, what loaded, next step: and then it goes straight into your real task without waiting.

When it fires

“set MasterMind up for this project” “init” “I want to build a booking site, start from scratch” “get me ready”

Or simply on your first real task in a project it hasn’t seen. You’ll see it in your terminal:

🧠 MasterMind ▸ setting up your project: will confirm what's wired
   └ init · detect stack → load field pack → ready

When it does not fire

  • You want to know what MasterMind can do. That’s help: a menu of every skill and when each one fires. The difference is direction: help explains MasterMind to you; init explains your project to MasterMind. If you’re not sure where to start, ask for help first.
  • The project is already set up. It checks, and skips itself rather than redoing work.
  • You want your repo scaffolded. Init configures knowledge, not your codebase. It won’t create files or edit your project on its own. That’s build, and it happens as part of an actual task.

What you get

A MasterMind that knows your framework and versions, matches your existing conventions instead of overwriting them, and loaded the right knowledge for both halves of a full-stack app.

At the very end it makes one small, entirely skippable offer: two optional preferences: a written report at the end of each work cycle, and a plan-first mode that shows you the plan before editing anything. Both are off unless you turn them on, and ignoring the question leaves them off.