← ClaudeAtlas

bld-sprint-initlisted

si: Take an app idea to a good-looking, working, deployed base. Design direction, Vite + React + shadcn scaffold, core screens, deploy. Use for "start a new app", "build me an app", "spin up a new project". Polish is /bld-sprint-refine.
mnguyenht/bld-package · ★ 1 · Web & Frontend · score 64
Install: claude install-skill mnguyenht/bld-package
# bld-sprint-init — idea → good-looking, deployed base One sprint from an app idea to a live, decent-looking base. This skill **orchestrates skills we already have** — don't reimplement them, invoke them. Scope is a strong BASE: real structure, on-brand look, deployed. Deep polish, animation, and craft passes belong to `/bld-sprint-refine`. Default stack: **Vite + React + TypeScript + Tailwind + shadcn/ui** (user's preference). Keep it beginner-friendly — explain approval-required commands first. Goal: make it look **as good as it can in the time budget the user picks** — not a deliberately bare base. The budget sets how far to push, not whether to care. ## Phase 0 — Ask the time budget (FIRST) Before anything, ask how much time/depth they want, and scale every phase to it. Use AskUserQuestion with these tiers: | Tier | Rough time | What it means | |------|-----------|---------------| | ⚡ **Quick** | ~10–15 min | Scaffold + MASTER.md + ONE polished core screen + deploy. Clean and on-brand, minimal scope. | | 🎯 **Standard** | ~30–45 min | All core screens, responsive, on-brand, light polish (hover/focus/transitions), deploy. Default. | | 🏗️ **Thorough** | ~60+ min | Standard + a first real polish pass (Emil-style micro-interactions, an impeccable audit), more screens/states. | Push quality to the ceiling of the chosen tier — a Quick app should still look genuinely good, just smaller in scope. The strict, exhaustive craft passes (full impeccable critique, review-animat