PRD template for coding agents

I'm building [product name], a [one-line description].

Target user: [who is this for]
Core problem: [what pain does this solve]
Key features I want in v1:
- [feature 1]
- [feature 2]
- [feature 3]

Tech stack: [e.g., Next.js, Supabase, Tailwind]

Based on this, generate a detailed PRD in structured JSON format that I can
feed to a coding agent. Include: overview, user stories, data models,
API endpoints, UI/page descriptions, auth requirements, and edge cases.

{
  "project": {
    "name": "YourFavGhostwriter",
    "description": "AI-powered LinkedIn ghostwriting tool that learns authentic writing voices",
    "tech_stack": ["Next.js", "Supabase", "Tailwind CSS", "Claude API"]
  },
  "features": [
    {
      "id": "F001",
      "name": "Voice Training",
      "priority": "P0",
      "description": "System ingests 10+ LinkedIn posts from user to learn their writing style",
      "user_story": "As a LinkedIn creator, I want the AI to match my tone so posts feel authentic",
      "acceptance_criteria": [
        "User can paste or import at least 10 past posts",
        "System extracts voice profile: tone, vocabulary, sentence structure, emoji usage",
        "Voice profile is stored and applied to all generated content"
      ],
      "data_model": {
        "table": "voice_profiles",
        "fields": {
          "id": "uuid, primary key",
          "user_id": "uuid, references users.id",
          "sample_posts": "jsonb, array of original posts",
          "voice_summary": "text, AI-generated style description",
          "created_at": "timestamp"
        }
      },
      "api_endpoints": [
        {
          "method": "POST",
          "path": "/api/voice/train",
          "body": "{ posts: string[] }",
          "response": "{ voice_profile_id: string, summary: string }"
        }
      ]
    }
  ],
  "pages": [
    {
      "route": "/dashboard",
      "description": "Main dashboard showing generated drafts and voice profile status",
      "components": ["DraftList", "VoiceProfileCard", "GenerateButton"],
      "auth_required": true
    }
  ],
  "auth": {
    "provider": "Supabase Auth",
    "methods": ["magic_link", "google_oauth"],
    "roles": ["free_user", "pro_user", "admin"]
  },
  "constraints": {
    "performance": "All pages load under 2s on 3G",
    "security": "Row Level Security on all Supabase tables",
    "api_limits": "Rate limit AI generation to 20 requests/hour for free tier"
  }
}

<prd>
  <project name="YourFavGhostwriter">
    <description>AI-powered LinkedIn ghostwriting tool that learns authentic writing voices</description>
    <tech_stack>
      <technology>Next.js</technology>
      <technology>Supabase</technology>
      <technology>Tailwind CSS</technology>
      <technology>Claude API</technology>
    </tech_stack>
  </project>

  <features>
    <feature id="F001" priority="P0">
      <name>Voice Training</name>
      <description>System ingests 10+ LinkedIn posts from user to learn their writing style</description>
      <user_story>As a LinkedIn creator, I want the AI to match my tone so posts feel authentic</user_story>
      <acceptance_criteria>
        <criterion>User can paste or import at least 10 past posts</criterion>
        <criterion>System extracts voice profile: tone, vocabulary, sentence structure</criterion>
        <criterion>Voice profile is stored and applied to all generated content</criterion>
      </acceptance_criteria>
      <data_model table="voice_profiles">
        <field name="id" type="uuid">primary key</field>
        <field name="user_id" type="uuid">references users.id</field>
        <field name="sample_posts" type="jsonb">array of original posts</field>
        <field name="voice_summary" type="text">AI-generated style description</field>
      </data_model>
    </feature>
  </features>
</prd>

You are a senior full-stack developer building a production-ready application.

Here is the complete Product Requirements Document (PRD) for the project:

<prd>
[paste your JSON/XML PRD here]
</prd>

INSTRUCTIONS:
1. Read the entire PRD before writing any code
2. Implement features in priority order (P0 first, then P1, then P2)
3. Follow the exact data models specified, do not add or remove fields
4. Match the API endpoints exactly as defined
5. For each page, implement all listed components
6. Apply all constraints (performance, security, rate limits)
7. Use the specified tech stack, do not substitute libraries
8. Add proper error handling and loading states for every user-facing feature
9. Write clean, typed code with clear file organization

Start with the project scaffold and database schema, then implement features
one by one. After each feature, confirm completion before moving to the next.

{
  "decisions": [
    { "decision": "Use server components by default, client components only when needed" },
    { "decision": "All API routes return consistent { data, error } response shape" },
    { "decision": "Use Supabase RLS instead of middleware for authorization" }
  ]
}