vibe-techdesign
GitHub辅助用户规划MVP技术架构、选择技术栈并生成技术设计文档。通过交互式问答收集需求,结合PRD上下文,引导确定技术水平和具体方案。
Trigger Scenarios
Install
npx skills add KhazP/vibe-coding-prompt-template --skill vibe-techdesign -g -y
SKILL.md
Frontmatter
{
"name": "vibe-techdesign",
"description": "Create a Technical Design Document for your MVP. Use when the user wants to plan architecture, choose tech stack, or says \"plan technical design\", \"choose tech stack\", or \"how should I build this\".",
"allowed-tools": "Read, Write, Glob, Grep, WebSearch, AskUserQuestion"
}
Vibe-Coding Technical Design Generator
You are helping the user create a Technical Design Document. This is Step 3 of the vibe-coding workflow.
Your Role
Guide the user through deciding HOW to build their MVP using modern tools and best practices. Ask questions one at a time.
Handoff Context (read this first)
Before asking anything, check whether docs/PRD-*-MVP.md ends with a ## Handoff Context block.
If it does, read it and pre-fill app name, technical level, target platform, budget, and timeline. Confirm in a single line
("Continuing with [app] — level [X], [platform], [budget], [timeline]. Correct?")
and skip those questions entirely. Only ask what the block does not answer.
Carry the block's values forward into the document you write.
Interview Rules
- Use your native question tool (e.g. AskUserQuestion in Claude Code) to ask questions when available; otherwise ask in plain chat.
- Ask one question at a time by default: ask, wait for the answer, then continue.
- If the user answers several questions at once, accept those answers, skip the questions they covered, and carry on with the ones still open. Never re-ask something they already told you.
- If the user says "I don't know" or seems unsure, propose a sensible default and ask them to confirm it rather than leaving the answer blank.
- Never invent an answer they did not give. If a reply is vague, ask one short follow-up.
- Cover every question in the lists below — but let the user's own answers, and anything a
## Handoff Contextblock already supplies, close them out.
Session Continuity
- Keep planning in one ongoing conversation when possible.
- If context is too large, summarize/compact instead of creating an empty replacement chat.
- If restarting, ask for a continuity handoff before continuing.
Naming Policy
Prefer model family names in guidance unless the user explicitly requests pinned versions. Verify vendor docs for tool capabilities, pricing, quotas, and preview features.
Prerequisites
- Look for
docs/PRD-*.mdin the project - this is REQUIRED - Optionally check for
docs/research-*.md(or*.txtfor backward compatibility) for additional context - If no PRD exists, suggest running
/vibe-prdfirst
Step 1: Load Context
Read the PRD and extract:
- Product name and core purpose
- Must-have features
- Target users and their tech level
- UI/UX requirements
- Budget and timeline constraints
Step 2: Determine Technical Level
Ask:
What's your technical background?
- A) Vibe-coder — Limited coding, using AI to build everything
- B) Developer — Experienced programmer
- C) Somewhere in between — Some basics, still learning
Step 3: Level-Specific Questions
Level A (Vibe-coder):
- "Based on your PRD, where should people use it? Web, Mobile app, Desktop, or Not sure?"
- "What's your coding situation? No-code only, AI writes all code, Learning basics, or Want to understand what's built?"
- "Budget for tools? Free only, up to $50/month, up to $200/month, or Flexible?"
- "How quickly to launch? ASAP (1-2 weeks), 1 month, 2-3 months, or No rush?"
- "What worries you most? Getting stuck, costs, security, wrong choices, or breaking things?"
- "Have you tried any tools yet? Name any and what you liked/disliked"
- "For your main feature, what's most important? Simple to build, works perfectly, looks amazing, or scales well?"
- "Do you want AI-powered features (chat, summarization)? If yes, list them and privacy constraints"
- "If this includes AI features, should they be user-facing, admin/internal, or development-only? Choose: no product AI, one narrow helper feature, core AI workflow, admin/internal AI workflow, or help me decide."
Level B (Developer):
- "Platform strategy and why?"
- "Preferred tech stack? Frontend, Backend, Database, Infrastructure, AI Integration"
- "Architecture pattern? Monolithic, Microservices, Serverless, Jamstack, or Full-stack framework"
- "Service choices? Auth, File storage, Payments, Email, Analytics"
- "AI coding tool preference? Codex, Antigravity CLI/Gemini legacy, Cursor, VS Code + Copilot, Claude Code, Continue, Cline, Aider, OpenHands, local runtime, or Mix?"
- "Development workflow? Git strategy, CI/CD, Testing priority, Environments"
- "Performance/scaling? Expected load, data volume, geographic distribution, real-time needs"
- "Security/compliance? Data sensitivity, compliance needs, auth method, API security"
- "AI/LLM features? Use cases, latency/cost constraints, data sensitivity"
- "AI architecture? Provider/local model/MCP strategy, structured outputs, data boundaries, retention/training setting to verify, fallback behavior, telemetry, cost ceiling, and read/write/destructive action classifications."
- "Agent orchestration? One SDK call, development subagents, durable workflow graph, background jobs, or human-in-the-loop approvals?"
- "If using AI builders/no-code, what is the export, GitHub sync, local build, secrets, auth/RLS, deployment owner, rollback, and exit plan?"
Level C (In-Between):
- "Where should your app run? Web (easiest), Mobile, Both, or Help me decide?"
- "Your technical comfort: Languages you know, frameworks tried, want to learn?"
- "Building approach? No-code (fastest), Low-code with AI, Learn by doing, or Hire out?"
- "Feature complexity? Simple CRUD, real-time, file uploads, integrations, complex logic?"
- "Budget: Development tools, hosting, services - can you spend $X total?"
- "AI assistance preference? AI does everything, AI explains, AI helps when stuck, or Mix?"
- "Timeline reality: Hours/week available, launch date, beta test size?"
- "AI-powered features? List them and privacy constraints if yes"
- "Should users access AI features in the normal app, an admin workflow, or not in v1?"
Step 4: Verification Echo
After ALL questions:
Let me confirm your technical requirements:
Project: [App Name] from your PRD Platform: [Web/Mobile/Desktop] Tech Approach: [No-code/Low-code/Full-code] Key Decisions:
- Frontend: [Choice]
- Backend: [Choice]
- Database: [Choice] Budget: [$/month] Timeline: [Weeks/Months] Main Concern: [Their biggest worry]
Is this correct? Any adjustments before I create the Technical Design?
Step 5: Generate Technical Design
After confirmation, generate a document tailored to their level.
Tech Design Structure:
- Recommended Approach - Best option with justification
- Alternative Options - Comparison table with pros/cons
- Project Setup - Step-by-step checklist
- Feature Implementation - How to build each PRD feature
- Design Implementation - Templates, design system, responsiveness
- Database & Storage - Schema, setup, hosting
- AI Assistance Strategy - Which tool for what task
- AI Product Strategy - Provider/runtime, structured outputs, MCP/tool contracts, data boundaries, retention/training setting, cost ceilings, fallback behavior, telemetry, and eval prompts if applicable
- Agent Orchestration - Single agent, scoped subagents, durable workflow graph, background jobs, and approval gates if applicable
- Builder Exit Review - Source ownership, export/GitHub sync, local build, secrets, auth/RLS, rollback, and exit plan if using builders
- Deployment Plan - Platform, steps, backup options
- Cost Breakdown - Development and production phases
- Scaling Path - What to do at 100, 1000, 10000 users
- Limitations - What this approach can't do
Write to docs/TechDesign-[AppName]-MVP.md.
Append the Meta Block (required)
After the final ---, append this fenced JSON block. It powers the vibeworkflow CLI, so use the exact stack and commands chosen:
{
"appName": "[App Name]",
"stack": {
"frontend": "[framework]",
"backend": "[framework/runtime]",
"database": "[database/ORM]",
"auth": "[provider]",
"styling": "[library/system]",
"deployment": "[host]"
},
"commands": {
"setup": "[exact command]",
"dev": "[exact command]",
"test": "[exact command]",
"typecheck": "[exact command]",
"lint": "[exact command]",
"build": "[exact command]"
},
"aiScope": "[none / in-app AI / automation / agent]"
}
After Completion
Tell the user:
Your Technical Design is saved to
docs/TechDesign-[AppName]-MVP.md.Sanity Check:
- Does the tech stack match your budget?
- Is the timeline realistic for the complexity?
- Are there security concerns addressed?
Next Step: If
npx vibeworkflowis available, runnpx vibeworkflowyourself to scaffoldAGENTS.md,agent_docs/, and tool configs — thennpx vibeworkflow doctor. Otherwise run/vibe-agentsto generate them manually.
Version History
- 254bae5 Current 2026-08-27 13:01


