Skip to content

golden-path-checklist.md

Latest commit

 

History

History
92 lines (73 loc) · 3.7 KB

golden-path-checklist.md

File metadata and controls

92 lines (73 loc) · 3.7 KB

Golden Path Integration Checklist

Use this checklist to validate the end-to-end workflow produces the expected files at each step.

Step 1: Deep Research

Input: User's idea description + part1-deepresearch.md prompt Output:

  • docs/research-[AppName].md exists (.txt accepted for backward compatibility)
  • Document includes: market analysis, competitors, technical recommendations, MVP feature prioritization

Handoff to Step 2: Research document or active chat session

Step 2: Product Requirements (PRD)

Input: Research output + part2-prd-mvp.md prompt Output:

  • docs/PRD-[AppName]-MVP.md exists
  • Document includes: product overview, target users, must-have features, success metrics, design direction

Handoff to Step 3: PRD document or active chat session

Step 3: Technical Design

Input: PRD output + part3-tech-design-mvp.md prompt Output:

  • docs/TechDesign-[AppName]-MVP.md exists
  • Document includes: tech stack, project structure, implementation approach, deployment plan, cost estimates

Handoff to Step 4: Technical Design document

Step 4: Agent Configuration

Input: PRD + Technical Design + part4-notes-for-agent.md Output:

  • AGENTS.md exists in project root
  • MEMORY.md exists in project root
  • REVIEW-CHECKLIST.md exists in project root
  • agent_docs/tech_stack.md exists and is populated
  • agent_docs/code_patterns.md exists and is populated
  • agent_docs/project_brief.md exists and is populated
  • agent_docs/product_requirements.md exists and is populated
  • agent_docs/testing.md exists and is populated
  • Tool-specific config exists based on user selection:
    • Claude Code: CLAUDE.md
    • Cursor: .cursor/rules/ or .cursorrules
    • Gemini CLI: GEMINI.md
    • VS Code + Copilot: .github/copilot-instructions.md

Handoff to Step 5: All files above in project root

Step 5: Build MVP

Input: All Step 4 outputs + user's coding environment Expected behavior:

  • Agent reads AGENTS.md first
  • Agent proposes a Phase 1 plan before coding
  • Agent builds one feature at a time
  • Agent runs tests/verification after each feature
  • Agent updates AGENTS.md current state after progress
  • Agent follows patterns in agent_docs/code_patterns.md

File contract summary

your-app/
├── docs/
│   ├── research-[AppName].md       ← Step 1 output
│   ├── PRD-[AppName]-MVP.md        ← Step 2 output
│   └── TechDesign-[AppName]-MVP.md ← Step 3 output
├── AGENTS.md                       ← Step 4 output (master plan)
├── MEMORY.md                       ← Step 4 output (session continuity)
├── REVIEW-CHECKLIST.md             ← Step 4 output
├── agent_docs/                     ← Step 4 output (detail docs)
│   ├── tech_stack.md
│   ├── code_patterns.md
│   ├── project_brief.md
│   ├── product_requirements.md
│   └── testing.md
├── [tool-specific configs]         ← Step 4 output
├── specs/                          ← Created during Step 5 (handoff artifacts)
└── src/                            ← Created during Step 5 (application code)

Claude skill discovery check

If using Claude Code skills, verify:

  • /vibe-research finds and reads docs/research-*.md (or .txt)
  • /vibe-prd finds and reads docs/research-*.md and writes docs/PRD-*.md
  • /vibe-techdesign finds and reads docs/PRD-*.md and writes docs/TechDesign-*.md
  • /vibe-agents finds docs/PRD-*.md and docs/TechDesign-*.md and generates all config files
  • /vibe-build finds AGENTS.md and agent_docs/ and starts the build loop