Przejdź do treści
Biblioteka skilli
AI i agenci · Automatyzacja

init

/init

Konfiguruje MasterMind w projekcie, który nie jest jeszcze przygotowany, przy pierwszej istotnej pracy lub gdy żaden pakiet dziedzinowy nie pasuje do wykrytego stosu.

Repozytorium
mastermind ↗
Licencja
MIT
Ostatnia zmiana repozytorium
08.10.2026

Ocena i bezpieczeństwo

100/ 100Klasa: dobry
Format25 / 25
Frontmatter z nazwą i opisem, poprawna nazwa, opis do 1024 znaków, nazwa zgodna z folderem.
Odkrywalność25 / 25
Opis mówi, kiedy użyć skilla, ma sensowną długość i jest w 3. osobie.
Treść35 / 35
Co najmniej 150 słów i do 500 linii, nagłówki, listy kroków, przykłady albo kod.
Higiena15 / 15
Bez zaślepek typu TODO, bez ścieżek Windows, z licencją.
Ryzykobrak sygnałów w skanie
Pliki w paczce
2
Skrypty
0
Komentarze HTML
0
Adresy URL w SKILL.md
0
Przestarzałość

Brak sygnałów. Repozytorium zmienione 08.10.2026, bez odwołań do wycofanych modeli i API.

Licencja

MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.

Co robi i kiedy

Konfiguruje MasterMind w projekcie, który nie jest jeszcze przygotowany, przy pierwszej istotnej pracy lub gdy żaden pakiet dziedzinowy nie pasuje do wykrytego stosu.

Wyzwalacze z opisu

t set up for yet, when the user says \, \, or when no field pack matches the project
Oryginalny opis (EN)

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.

Jak działa

  1. Keep it short (the 1 rule)
  2. Path A: the project already has code
  3. Path B: greenfield (empty folder / nothing built yet)
  4. The first result comes before setup finishes
  5. Show the plan before you write the pack
  6. Always end with a short report
  7. Optional preferences: offer once, then remember
  8. Gotchas
Sekcje z nagłówków SKILL.md, w oryginale · 126 linii, 1 328 słów
Podgląd SKILL.md
---
name: init
description: "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."
---

# Initialize: get MasterMind ready for this project, fast

The goal: a new user goes from "installed" to "the brain is set up for my stack and knows what I'm
building" in a **few quick steps**: a handoff, not an intake form. Detect what you can; ask the rest.
Run **once per project**; if this project already built its field, you're set: skip this.

**Nothing ships pre-baked.** A fresh install carries only the engine and `engineering/fields/_template/`
no field, because a pack tuned to someone else's stack is worse than none (dead weight that misleads).
So init's core job is to **build this project's field from the template**, tuned to the real stack. This
is where the measured quality lift comes from; without it you still apply solid general judgment, but the
domain pack is the difference.

## Keep it short (the #1 rule)
**Detect before asking: read the answers out of the project, and make the technical calls yourself.**
A developed project needs **zero questions**: continue with its own stack. A greenfield project needs **one
open question** ("what do you want to build?"), answered in their own words. The whole thing should feel
like 20 seconds, then you're working.

## Path A: the project already has code
1. **Detect the stack**: `package.json`/lockfile, configs, framework + versions, DB, test runner, folder
   shape (`core/agent-loop.md`). This is free, read it rather than asking.
2. **Set up its field:**
   - **This project already built a field that fits** → load it. Done. (Nothing else ships, so this only
     happens on a re-run of a project that was set up before.)
   - **Otherwise: build one** from `engineering/fields/_template/`, tuned to the detected stack: a
     **one-time bootstrap** (`levelup bootstrap`), stated as a trade-off, not a menu: *"a few minutes now
     makes me much sharper on your stack: its real defaults, pitfalls, and review rules. It's once per
     field, then reused for every task."* Build it to fit **one real stack**, and keep it lean, tailoring
     **prunes as much as it adds** (`active-field.md`). **Fill the pack's "Where things are" section**
     from what step 1 already detected: the half-dozen entry-point pointers (components, routes, state,
     tests + run command, config, the one file that explains the architecture fastest). This is the
     narrow exception to "skip the directory listing" below: a full tree is re-derivable noise, but
     these few pointers are what stop every future task from re-running discovery. Then point
     `active-field.md` at it and regenerate the router **from the brain you are installed into**:
     `node .mastermind/scripts/build-router.mjs` for a per-project brain, the default. Under
     `--shared` the brain is the one in your home directory, so run that same script from there.
     A bare `scripts/…` path is the project's own directory and fails with MODULE_NOT_FOUND.
3. **Fill the project brief** (`.mastermind/brief.md`). An isolated install seeds it; under
   `--shared` there is no project brain, so create the file yourself at that path. The field pack holds the
   stack; the brief holds *this* project: what it is, the words it uses for its own ideas, what breaks
   and who feels it, what cannot be undone here, and how it ships. Most of it comes from step 1. Ask
   the user only for what the code cannot tell you, which is usually who feels an outage and which
   actions are irreversible. **Keep it under about 40 lines**: it is read every session. This is what
   stops a change that is technically correct and operationally wrong.
4. **Respect the project**: match its conventions and keep the stack it already chose
   (`stack-defaults.md`: the project wins).

## Path B: greenfield (empty folder / nothing built yet)
Nothing to detect, so ask **one open question, in the user's own words rather than a menu of stack options:**

> *"What do you want to build?"*

Let them describe it freely: a product, an app, a feature, **technical or not**. Then read their answer:
- **They named the tech** (a framework, language, database) → use it.
- **They only described the product** → assume they may not be an engineer, and **make the stack call for
  them.** *You* choose the architecture and tech stack that best fit what they described. That's the
  prime directive (decide for them), then state the choice with a one-line *why* and **proceed without
  pausing**: *"I'll build this as Next.js + Postgres: best fit for a realtime habit tracker."* (If they
  later want something different, they'll say so.)

Then set up the field pack(s) for that stack, **load or bootstrap all it needs** (a real app needs
**frontend + backend**: UI *and* auth/tenancy/data/jobs). Offer to scaffold the starting point
(repo/toolchain/env), then hand to `build` / `architect` for the first slice.

## The first result comes before setup finishes
If the user is eager to build *now*, start the first vertical slice on general judgment and run
pack-bootstrap **as an after-first-slice / background step**: a multi-minute research detour waits its
turn behind their first working result.

## Show the plan before you write the pack

Building a field takes minutes of research and lands real files, so let the user see it as a proposal
first: a few lines, not a document:

```text
Detected: Next.js 15 · React 19 · TypeScript · Postgres/Prisma · Vitest
I'll build a `frontend` field with: stack defaults for this exact combination ·
review rules for React 19 · an empty lessons file that fills as we work.
Starting now: say the word if the stack reading is off.
```

State it and proceed; this announces rather than asks. It costs one line and catches the one failure
worth catching: a wrong stack reading becoming a pack that's subtly wrong about everything.

**Store only what can't be re-derived**: plus the pack's few "Where things are" entry-point pointers
(step 2's named exception). Skip the full directory listing and the dependency list, the model
can read those any time. Keep the pitfalls, the non-obvious conventions, and the decisions someone
would otherwise have to rediscover.

## Always end with a short report
Both paths finish with a **~5-line** report so the setup is visible and the user knows what's ready:

```
▸ MasterMind ready
  stack:       Next.js 15 · React 19 · Postgres/Prisma   (detected)
  fields:      frontend + backend (built from _template for your stack)
  conventions: matched your existing style
  next:        tell me what to build, I'll run the build loop.
```

Then proceed straight into their actual task: the report is a line you pass through, not a checkpoint.

## Optional preferences: offer once, then remember
After the ready report, offer these once: both skippable, both default **off**:

> *"Two optional preferences, both off by default. I'll remember your answers:*
> *(1) a short written **report** at the end of each build/QA cycle? (markdown / html)*
> *(2) **plan-first**: on bigger tasks, should I show the plan and wait for your OK before editing?"*

Record the answers in **`.mastermind/prefs.md`**, seeded by an isolated install and created by you
under `--shared`, one line each. Both preferences are defined by
their implementations, **`skills/build/SKILL.md`** (plan-first gate) and **`skills/report/SKILL.md`**
(`cycle-report` values); read those for the exact keys and accepted values rather than restating them here.
Ask once and let it go: a shrug or silence defaults both `off` and you move straight on. They can change
either anytime ("reports on", "plan first from now on"). (Keep it to this single light offer: one ask,
then work.)

## Gotchas
- **Run once per project**: check for a matching field pack first, and load it if one is already there.
- **One open question, kept open-ended**: greenfield asks only "what do you want to build?"; if the user
  isn't technical, *you* pick the stack (decide for them), state it, and move, the choice is yours to make.
- **Full-stack means both packs**: the classic failure is a polished UI on an unguarded backend; load
  the backend pack whenever auth/data/tenancy/billing are in scope.
- This sets up *knowledge*, not your repo's config, repo edits happen only as part of doing the task.
Pliki w paczce (2)
  • ABOUT.md 4 KB
  • SKILL.md 8 KB

Wersje

RepozytoriumWynikLicencja
Ta strona · mehrad-dm/mastermind100MIT
CodaColaLight/nightmill100Apache-2.0
userKael/rastro100MIT
elalfner/relevo100MIT
abatyuk/okf100MIT
duronext/jig100MIT
sh5623/e2e-rail100MIT
Calvin-LLC/skills100MIT
arxdsilva/ai-loops100MIT

Migawka bazy z 13.11.2026 · wersja paczki 6886839f099b (skrót SHA-256 całej paczki)

Pobrania w ostatnich 30 dniach: 0 · ranking

Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.

Podobne skille

  • 100help
    Wyjaśnia, co potrafi MasterMind i jak nim sterować: lista poleceń, skilli i opcji dla osób, które nie wiedzą, od czego zacząć.
  • 100lint
    Sprawdza pliki instrukcji MasterMind (kernel, core, skille, agenci) pod kątem rozdęcia, powtórzonych reguł i sprzeczności między warstwami, raportuje ustalenia z file:line i czeka na akceptację zmian.
  • 100init-pipeline
    Konfiguruje potok subagent-system w projekcie: tworzy PIPELINE_CONFIG.md i szkielet punktów kontrolnych.
  • 100project-init
    Konfiguruje zwięzły kontekst projektu dla Claude, Codex i Grok po przejrzeniu repozytorium, na pierwszej konfiguracji lub po sklonowaniu.
  • 96hub-init
    Konfiguruje Session Hub dla projektu i otwiera jego interfejs webowy, uruchamiając wspólny serwer, jeśli nie działa.
  • 92init-project
    Pomaga przygotować nowy lub istniejący projekt do pracy wspomaganej AI.

Uwagi i zgłoszenia

Wiadomość trafi do Bartosza Sokolińskiego. Adres e-mail jest potrzebny tylko do odpowiedzi. Szczegóły w polityce prywatności.