Help us improve
Share bugs, ideas, or general feedback.
From pm-vibe-coding
Plan an AI-assisted prototyping session: scope the MVP, pick the right AI coding tool, define the build sequence, and estimate effort for a solo PM-builder.
npx claudepluginhub tarunccet/pm-skills --plugin pm-vibe-codingHow this skill is triggered — by the user, by Claude, or both
Slash command
/pm-vibe-coding:prototype-planThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
| Input | Required? | Example |
Use this skill when the user asks to "vibe code this", "build this with AI", "help me use Cursor/v0/Bolt to build this", "vibe coding from my PRD", "how do I code this with AI", "turn this spec into code", or wants guidance on using AI coding tools (Cursor, GitHub Copilot, v0, Bolt, Lovable, Claude Artifacts) to prototype or build a feature from a product spec. This is a coaching skill — it helps the PM get the most out of AI coding tools, not write code directly.
Generates phased roadmaps for full application builds and executes autonomously—committing, deploying, testing phase-by-phase. Modes: plan, start, resume, status.
Generates structured PRDs for complex Replit apps with executive summaries, user personas, tech stacks, architecture diagrams, and acceptance criteria. Use for multi-feature projects needing specs for AI-powered autonomous builds.
Share bugs, ideas, or general feedback.
| Input | Required? | Example |
|---|---|---|
| What you want to build | ✅ Required | "A waitlist page with referral tracking and email capture" |
| Primary purpose of the prototype | ✅ Required | Validate demand before building the full product |
| Your technical comfort level | 🟡 Recommended | "I can read code but not write it" / "Comfortable with no-code" |
| Deadline or time constraint | ⚪ Optional | "Need a demo-ready version in 3 days" |
Don't have everything? Start anyway — the skill will work with what you provide and flag where richer input would improve the output.
Create a structured plan for building a product prototype using AI coding assistants. Covers tool selection, MVP scoping, build sequence (data model → backend → frontend → deploy), and effort estimation for a solo PM-builder. This prevents the common trap of starting to code without a plan and ending up with a broken half-built prototype.
AI coding assistants (Cursor, Replit Agent, GitHub Copilot, Bolt.new, v0.dev, etc.) each have different strengths. Cursor excels at complex full-stack apps with a professional codebase. Replit Agent is fastest for quick standalone apps. Bolt.new/v0.dev are best for frontend-only. Choosing the wrong tool adds hours of friction.
The build sequence matters: data model → backend API → frontend → auth → deployment. Building frontend first and then trying to fit the data model is the #1 cause of PM-builder rewrites. Think of it like building a house: foundation before walls, walls before roof.
Understanding realistic effort estimates is critical. AI coding speeds up development by 3-5x for experienced developers, but for PM-builders who are reviewing and guiding AI output (not just accepting it), sessions tend to run longer due to debugging and iteration. The estimates here account for that reality.
You are helping a PM-builder plan a prototype build for $ARGUMENTS.
Generate a prototype plan covering:
Recommend the best tool(s) based on the project's needs:
| Tool | Best For | Strengths | Limitations |
|---|---|---|---|
| Bolt.new | Frontend-only, React components, no backend | Instant preview, great UI generation | No persistent backend |
| v0.dev | UI component generation, Shadcn UI | Beautiful components fast | Frontend only |
| Replit Agent | Quick full-stack apps, demos | Zero setup, DB included, instant deploy | Less control, not production-grade |
| Cursor | Complex full-stack apps, existing codebases | Fine-grained control, file editing | Requires setup, slower start |
| GitHub Copilot | Existing VS Code projects, code completion | IDE-native, great for iteration | Not great for greenfield from scratch |
| Claude Code | Complex logic, architecture, debugging, refactoring | Strong reasoning, great for debugging | Terminal-based, steeper learning curve |
| Windsurf | Full-stack apps with agentic flows | Good at multi-file changes | Newer, less battle-tested |
Explain specifically why your recommendation(s) fit this project. Consider:
Break the build into sessions (assuming 2-4 hour focused sessions):
Session 1: Foundation & Data Model
Session 2: Backend / API Layer
Session 3: Core Frontend
Session 4: Authentication
Session 5: Polish & Deploy
/deploy-check)| Session | Core Tasks | Estimated Hours | Recommended AI Tool |
|---|---|---|---|
| 1: Foundation | Project setup, schema, DB connection | 2–3h | Cursor or Replit |
| 2: Backend | API endpoints, validation, error handling | 3–4h | Cursor or Claude Code |
| 3: Frontend | Core flow UI, API integration, mobile | 3–4h | Cursor + v0.dev for components |
| 4: Auth | Sign-up, login, route protection | 2–3h | Cursor |
| 5: Polish & Deploy | UX fixes, env vars, deploy, test | 2–3h | Cursor + Vercel |
| Total | 12–17 hours |
Adjust estimates based on: complexity of data model (+/- 2h), number of external API integrations (+2h each), custom UI requirements (+2h), and your familiarity with the tech stack (+/- 3h).
Identify the top 3-5 risks for this specific project:
Common risks for PM-builders:
Before starting Session 1, verify:
A structured markdown plan ready to execute, with all 6 sections filled in for the specific project.