Skip to main content

setup

Conversational onboarding that generates a personalized design knowledge system. Asks about your design practice, then creates identity, methodology, MOCs, and processing pipeline.

Informações da origem

Repositório
codejunkie99/design-brain
Última atividade na origem
18 de fevereiro de 2026 às 13:15
Idioma detectado do SKILL.md
inglês
Estrelas
2
Forks
0

Opções de instalação

Por padrão, está selecionado o prompt que primeiro revisa a origem. Você pode mudar para um comando direto ou baixar uma cópia local.

Revise os arquivos de origem

Leia o SKILL.md e os arquivos complementares exibidos pelo SkillsMP antes de decidir se vai instalar.

Exibindo SKILL.md

SKILL.md
Instruções da origem · Visualização somente leitura
name
setup
description
Conversational onboarding that generates a personalized design knowledge system. Asks about your design practice, then creates identity, methodology, MOCs, and processing pipeline.
allowed-tools
["Bash","Read","Write","Edit","Glob","Grep","AskUserQuestion"]
# Design Brain Setup Generate a personalized design knowledge system through conversation. ## When to use - First time setting up design-brain in a workspace - Re-initializing after changing design focus - Adding to a workspace that already has `.design-brain/` from CLI captures ## 4-Phase Flow ### Phase 1: Detect Check the current state of the workspace: ```bash ls -la .design-brain/ 2>/dev/null ``` Determine: - Does `.design-brain/` exist? - Are there existing projects/captures? - Are knowledge system files (self/, notes/, ops/) already present? Report findings to the user: - "Found existing design brain with N projects and M inspirations" or - "No design brain found, starting fresh" ### Phase 2: Understand Ask 2-3 questions about the user's design practice. Use AskUserQuestion with multiple choice. **Question 1: Design domains** "What design domains do you work in?" Options (multiSelect: true): - Web product design (SaaS, dashboards, landing pages) - Brand & visual identity (logos, guidelines, color systems) - Mobile & native apps - Design systems engineering (tokens, components, Figma-to-code) **Question 2: Primary focus** "What's your main goal with design-brain?" Options: - Competitive analysis (study what others build) - Design system development (extract patterns into reusable systems) - Inspiration library (collect and organize reference material) - Cross-project knowledge (track decisions and principles across projects) **Question 3 (optional, based on Q2):** If design system development or cross-project knowledge: "What tools and frameworks do you primarily use?" Options (multiSelect: true): - Tailwind CSS - Figma - Storybook - Custom design system ### Phase 3: Generate Based on answers, run the following commands: 1. Initialize the brain if not present: ```bash design-brain-memory init --root . ``` 2. Generate identity file at `.design-brain/self/identity.md`: - Include selected domains - Include focus description - Include tools if provided 3. Generate methodology at `.design-brain/self/methodology.md` 4. Generate MOC hub at `.design-brain/notes/mocs/hub.md` 5. Generate domain MOCs for each selected domain at `.design-brain/notes/mocs/<domain>.md` 6. Generate maintenance file at `.design-brain/ops/maintenance.md` 7. If existing captures exist, update MOC hub with project links **Important:** Use the `generateIdentity`, `generateMethodology`, `generateMocHub`, `generateDomainMoc`, and `generateMaintenance` functions from the `design-brain-memory` package. These are available via: ```bash design-brain-memory reindex --root . ``` This will auto-generate all knowledge system files with sensible defaults. After reindex, update the identity file with the user's specific answers. ### Phase 4: Validate Verify all pieces exist: ```bash ls .design-brain/self/identity.md ls .design-brain/self/methodology.md ls .design-brain/notes/mocs/hub.md ls .design-brain/notes/mocs/color-systems.md ls .design-brain/notes/mocs/typography.md ls .design-brain/notes/mocs/components.md ls .design-brain/notes/mocs/layout.md ls .design-brain/notes/mocs/motion.md ls .design-brain/notes/mocs/brand.md ls .design-brain/ops/maintenance.md ls .design-brain/notes/patterns/ ls .design-brain/notes/decisions/ ls .design-brain/notes/principles/ ls .design-brain/ops/queue/ ls .design-brain/ops/sessions/ ``` Report results: - "Knowledge system ready! Created N files across self/, notes/, ops/" - List the key files created - Suggest next steps: "Run `/db-capture` to ingest your first inspiration, then `/db-extract` to generate pattern notes" ## Output After setup completes, the workspace should have: ``` .design-brain/ ├── self/ │ ├── identity.md # Personalized to user's answers │ └── methodology.md # Standard pipeline description ├── notes/ │ ├── patterns/ # Empty, ready for extraction │ ├── decisions/ # Empty, ready for decisions │ ├── principles/ # Empty, ready for principles │ └── mocs/ │ ├── hub.md # Links to all domain MOCs and projects │ ├── color-systems.md │ ├── typography.md │ ├── components.md │ ├── layout.md │ ├── motion.md │ └── brand.md ├── ops/ │ ├── queue/ # Empty, ready for pipeline tasks │ ├── sessions/ # Empty, ready for session logs │ └── maintenance.md # Initial health status ├── projects/ # Existing or empty ├── graph/ # Existing or empty └── database.json # Existing or fresh ```
Ver no GitHub