Project Initialization Skill
Activate this skill when bootstrapping a new project from scratch, establishing architectural baselines, or executing /init-project.
Operational Directive: Never write code before establishing product scope, technical requirements, and developer expertise.
1. Product Discovery Questionnaire
Before recommending or installing any framework, gather critical product invariants:
- Target Product & Core Value Proposition: What is being built, what core problem does it solve, and who are the primary users?
- Platform & Modality: Public SaaS, internal dashboard, mobile API, desktop application, or developer CLI?
- Scale & Traffic Expectations: Prototype / MVP, internal team tool (<100 users), or high-throughput multi-tenant production system?
- Data Model & Concurrency: Relational data (PostgreSQL/MySQL), document-based (MongoDB), key-value cache (Redis), or embedded (SQLite)?
- Deployment & Runtime Constraints: Serverless (Vercel/Cloudflare Workers), containerized cluster (Docker/Kubernetes), single VPS, or static hosting?
2. Developer Expertise Matrix
Assess the developer's familiarity level (Beginner, Intermediate, Advanced, Expert):
- Languages: TypeScript, Python, Go, Rust, C#, PHP, Java
- Frontend Frameworks: Next.js (App Router), React (SPA), Vue / Nuxt, SvelteKit, Vanilla HTML/CSS
- Backend Frameworks: Express / Fastify, NestJS, Django / FastAPI, Go Gin / Echo, ASP.NET Core, Laravel
- Databases & ORMs: Prisma, Drizzle, TypeORM, SQLAlchemy, Django ORM, Entity Framework Core, Raw SQL
Rule: Never recommend a complex, unfamiliar stack for an urgent MVP if a well-understood, robust alternative achieves the goal in half the time.
3. Technology Stack Recommendation Matrix
| Project Archetype | Recommended Stack | Key Rationale |
|---|---|---|
| Full-Stack SaaS / Web App | Next.js (App Router) + TypeScript + Tailwind CSS + PostgreSQL + Prisma/Drizzle | Unified language, fast server-side rendering, robust ORM, rapid UI iterations |
| High-Performance API Service | Go (net/http + Gin) OR FastAPI (Python) + PostgreSQL + Redis | High concurrency, low memory footprint, strict typing, auto-generated OpenAPI |
| Data Science / ML Backend | Python + FastAPI + PyTorch / Scikit-learn + PostgreSQL + Celery/Redis | Native ML ecosystem support, async performance, robust data processing |
| Enterprise Internal Tool | Next.js + React + shadcn/ui + PostgreSQL + Supabase / Auth.js | Rapid scaffolding, enterprise UI primitives, integrated authentication |
| Fast Developer CLI Tool | Node.js (TypeScript + Commander + Picocolors) OR Rust (Clap) | Zero-config distribution via npx, fast startup, deterministic execution |
| Micro-Service / Embedded | Go OR Rust + SQLite / PostgreSQL | Zero runtime dependencies, single static binary, instant cold start |
4. Initialization Output & Memory Scaffolding
Upon stack confirmation, generate the baseline architecture:
- Initialize
.ai/memory directory with:INDEX.md: Top-level memory index.CONSTITUTION.md: Project non-negotiables, intensity mode, and coding standards.STACK.md: Confirmed frameworks, runtimes, and dependencies.ARCHITECTURE.md: High-level system topology and boundaries.DECISIONS.md: Initial ADR-001 (Technology Stack & Baseline Architecture).
- Synchronize multi-agent adapter rules (
CLAUDE.md,.cursorrules,AGENTS.md,GEMINI.md).