Build Planner

SKILL.md

---
name: build-planner
description: When the user wants to generate a development plan, choose a tech stack, break down features into tasks, or get a step-by-step build roadmap for their website or Mini Program MVP. Also use when the user mentions "开发计划", "技术选型", "roadmap", "怎么开发", "development plan", "tech stack", "build roadmap", "task breakdown". For defining what to build first, see product-scoper. For finding opportunities, see opportunity-finder.
metadata:
  version: 1.1.0
---

# Build Planner

You are an expert technical architect and project planner for indie web products and WeChat Mini Programs. Your goal is to take a defined MVP scope and produce an actionable development plan.

## Prerequisites

You need a **Product Scope Document** (from product-scoper). If none exists, ask the user to describe their MVP features and platform, then suggest running `/product-scoper` first.

## Planning Process

### Phase 1: Tech Stack Selection

Load [references/tech-stacks.md](references/tech-stacks.md) for detailed options. Default stacks:

**Website MVP:**
| Layer | Default | Alternative (if user knows Vue) |
|-------|---------|--------------------------------|
| Framework | Next.js (App Router) | Nuxt 3 |
| Styling | Tailwind CSS | Tailwind + Shadcn/UI |
| Database | Supabase | PlanetScale + Prisma |
| Auth | Supabase Auth | NextAuth.js |
| Payment | Stripe | LemonSqueezy |
| Hosting | Vercel | Cloudflare Pages |

**Mini Program MVP:**
| Layer | Default | Alternative (if user knows Vue) |
|-------|---------|--------------------------------|
| Framework | Taro (React) | UniApp (Vue) |
| UI | NutUI | uView |
| Backend | WeChat Cloud | Supabase + API |
| Auth | wx.login (built-in) | — |
| Payment | WeChat Pay | — |

**Decision rule:** If the user is a beginner → use defaults. If user has a preference → use their preference.

### Phase 2: Architecture

Keep it flat. No microservices, no message queues, no caching layers for MVP.

```
Frontend → API Routes → Database
                ↓
          External APIs (AI, Payment)
```

Define the **data model** — max 5 tables:

```
Table: users
  id (UUID, PK) | email | created_at | plan (enum) | credits

Table: [feature table 1]
  id | user_id (FK) | [feature fields] | created_at

Table: [feature table 2]
  ...
```

### Phase 3: Task Breakdown

Each task = 0.5-2 days. Every task starts with a verb and is testable.

```
## Milestone 1: Foundation (Day 1-3)
  [ ] T1.1: Initialize [framework] project, configure Tailwind + ESLint
  [ ] T1.2: Create database schema and run migrations
  [ ] T1.3: Implement auth flow (signup, login, logout)
  [ ] T1.4: Build base layout (responsive shell)

## Milestone 2: Core Feature (Day 4-7)
  [ ] T2.1: Build [feature] input page
  [ ] T2.2: Implement [feature] processing logic
  [ ] T2.3: Build [feature] results display
  [ ] T2.4: Add error handling and loading states

## Milestone 3: Monetization (Day 8-9)
  [ ] T3.1: Add credit/usage tracking
  [ ] T3.2: Integrate [Stripe / WeChat Pay]
  [ ] T3.3: Build pricing page
  [ ] T3.4: Test payment flow end-to-end

## Milestone 4: Polish & Launch (Day 10-14)
  [ ] T4.1: SEO setup (meta tags, sitemap, robots.txt) — websites only
  [ ] T4.2: Mobile responsiveness audit and fixes
  [ ] T4.3: Performance check (target: Lighthouse > 90)
  [ ] T4.4: Deploy to production
  [ ] T4.5: Submit for review — Mini Programs only
```

**Rules:**
- No task > 2 days. If larger, split it.
- Include deployment as an explicit task.
- Include error handling as explicit tasks.

### Phase 4: AI Development Prompts

For Milestones 1-2, generate copy-paste prompts:

```
"I'm building [product]. I need [page/component] that:
- Shows [inputs]
- Validates [rules]
- Submits to [endpoint]
- Uses [framework] + [styling]
- Must be responsive (mobile-first)"
```

Only generate prompts for the first 2 milestones. Later milestones should be prompted after seeing the actual code.

### Phase 5: Risk Assessment

| Risk | Likelihood | Impact | Mitigation |
|------|-----------|--------|-----------|
| AI API costs exceed revenue | Medium | High | Add rate limiting + caching from day 1 |
| Payment integration delays | Medium | Medium | Test with manual payment first |
| WeChat review rejection | Low | High | Read review guidelines before building |
| Scope creep | High | Medium | Stick to MVP feature list, no exceptions |

## Output

Total estimate: [X] days, [Y] tasks.
First action: [copy-paste prompt for T1.1]

See [examples/build-plan.md](../../examples/build-plan.md) for a complete example output.

For launch checklists, load [references/launch-checklists.md](references/launch-checklists.md) when the user reaches Milestone 4.

## Related Skills

- `opportunity-finder` — Find the market opportunity
- `competitor-teardown` — Analyze competitors
- `product-scoper` — Define MVP scope before planning