Create Prd

Create PRD: Generate Product Requirements Document

Overview

Generate a comprehensive Product Requirements Document (PRD) through strict context gathering, API research, and structured documentation. This command produces a production-ready PRD serving as the single source of truth for implementation.

Output File: $ARGUMENTS (default: PRD.md in project root)

---

⚠️ CRITICAL INSTRUCTIONS

DO NOT CREATE THE PRD UNTIL YOU HAVE COMPLETE CONTEXT

Workflow:

1. Discover project type and scope through targeted questions
2. Research external APIs/SDKs if applicable (mandatory, not optional)
3. Gather all required information - ask questions until complete
4. Confirm with user before writing
5. Write PRD using appropriate template
6. Backup existing file if one exists

Principles:

• ❌ No assumptions - ask instead of guessing
• ❌ No placeholder sections - complete or explicitly skipped by user
• ✅ Complete context first - gather all info before writing
• ✅ Research external dependencies - understand APIs before documenting

---

STEP 1: PROJECT TYPE & CONTEXT DISCOVERY

Required Questions - Ask the user:

1. Project Type: AI Agent / Web App / API/Service / Library/Package / CLI Tool / Desktop App / Mobile App / Other
2. Core Functionality: In 2-3 sentences, what does this product do?
3. Target Users: Who will use this? Technical level?
4. MVP Scope: What must be in first version vs. later?
5. Technology Preferences: Required/preferred technologies?
6. External Dependencies: External APIs, services, or SDKs?
7. Deployment Context: How deployed/distributed?
8. Existing Documentation: Related docs or research?

Do not proceed without:

• ✓ What you're building (type + core function)
• ✓ Who it's for (target users)
• ✓ What's in scope for MVP
• ✓ What external dependencies exist

---

STEP 2: API RESEARCH (IF APPLICABLE)

Mandatory if external APIs or unfamiliar SDKs involved.

Identify & Determine Research Needs

List all external APIs/SDKs, then ask user for each:

This project integrates with [API name].

Options:
1. Do you have existing research documents? (provide paths)
2. Should I research [API name] before creating PRD?
3. Skip research and document based on current knowledge (not recommended for unfamiliar APIs)

Your preference?

Conduct Research

For each API requiring research:

1. Execute: /explore-api [API name]
2. Read: .agents/research/[api]-research.md
3. Verify: POC tests passed

Critical Verification Requirements

For EVERY external API/service, research MUST verify:

1. Feasibility: Is what we want to achieve technically possible with this API?

• Does the API provide the required features/endpoints?
• Are there version/compatibility constraints?
• Are there known limitations that block our use case?

1. Allowability: Is what we want to achieve permitted by the service?

• Rate limits sufficient for our needs?
• API tier/plan supports required features?
• Terms of Service allow our intended use?
• Geographic/regulatory restrictions?

1. Setup Requirements: What external setup is needed?

• Account creation/API key generation steps
• Dashboard/console configuration
• Webhooks or callback URL setup
• OAuth app registration
• Service-specific prerequisites

Do not proceed without:

• ✓ User confirmation on research approach
• ✓ Completed API research (if chosen)
• ✓ Research documents read (if provided)
• ✓ Feasibility verified (we CAN do what we want)
• ✓ Allowability verified (we're ALLOWED to do what we want)
• ✓ Setup requirements documented (we know HOW to configure it)

---

STEP 3: GATHER COMPLETE REQUIREMENTS

Information Gathering Strategy

For each template section:

1. Check conversation history for details
2. Ask specific questions for missing info
3. Confirm understanding before next section
4. Mark sections to skip (with explicit confirmation)

Required Information Checklist

Business Context:

• [ ] Product mission and value proposition
• [ ] Target user personas and needs
• [ ] MVP success criteria

Technical Context:

• [ ] Technology stack with versions
• [ ] Architecture approach and patterns
• [ ] External API integrations (with research if applicable)
• [ ] Security and configuration

Scope Definition:

• [ ] In-scope features for MVP
• [ ] Explicit out-of-scope items
• [ ] User stories with examples
• [ ] Implementation phases with parallelization strategy

Parallel Execution Design:

• [ ] Workstream boundaries identified
• [ ] Interface contracts between teams defined
• [ ] Shared context and dependencies mapped
• [ ] Integration checkpoints planned
• [ ] Mock/stub strategies for blocked work

Risk Management:

• [ ] 3-5 key risks identified
• [ ] Mitigation strategy per risk
• [ ] API integration risks (if applicable)

Example Questions (Adapt to Project Type)

AI Agents: Tools/capabilities? Registration pattern? Conversation flow? Result presentation? Web Apps: User workflows? Frontend framework? API design? Auth requirements? APIs/Services: Endpoints? Request/response formats? Rate limiting? Versioning? Libraries: Public API surface? Installation? Usage examples? Dependencies? Universal: Error handling? Logging? Tests (unit/integration/e2e)? Documentation?

Parallel Execution Planning:

• Can frontend and backend be built concurrently?
• What shared interfaces/contracts need to be defined first?
• Which components have no dependencies and can start immediately?
• What mock/stub strategies enable parallel development?
• Where are the integration points that require synchronization?

---

STEP 4: FILE HANDLING & PRD CREATION

Backup Existing PRD

If PRD exists:

mv PRD.md PRD.backup-$(date +%Y%m%d-%H%M%S).md

Confirm Before Writing

Ready to create PRD:
- Project Type: [type]
- Target File: [path]
- API Research: [completed/skipped/N/A]
- Sections: [number]
- Skipped: [list or "none"]

Proceed? (y/n)

---

PRD TEMPLATE STRUCTURE

Follow this hierarchical structure. Adapt sections based on project type.

Section Format

Each section below shows:

• Req: Required/Optional
• Format: Markdown syntax
• Examples: For different project types (AI Agent, Web App, API, Library)

---

# Product Requirements Document: [Product Name]

Req: Yes | Format: H1 with "Product Requirements Document: " prefix

---

## Executive Summary

Req: Yes | Format: H2, 2-3 paragraphs

[Product Name] - [One-liner]

• AI Agent: **Paddy** is an AI agent enabling Obsidian users to interact with vaults using natural language.
• Web App: **TaskFlow** is a collaborative PM app helping remote teams coordinate through task boards and async video.
• API: **PaymentHub** is a unified payment API abstracting multiple providers behind one interface.

MVP Goal: [One sentence]

• AI Agent: Enable vault query/management through natural language with user's choice of LLM.
• Web App: Allow teams to create task boards and share video updates without real-time presence.

---

## Mission

Req: Yes | Format: H2, 1-2 sentences

### Core Principles - Numbered list with bold names

1. **[Principle]** - [Explanation]
2. **[Principle]** - [Explanation]
3. **[Principle]** - [Explanation]

---

## Target Users

Req: Yes | Format: H2 with subsections

[User Group] who want to:

• Bullet list of goals

Technical Comfort Level: [Description]

---

## MVP Scope

Req: Yes | Format: H2 with H3 subsections

### In Scope - Categorized bullets:

• Core Functionality: User-facing features
• Technical: Implementation details
• Integration: Third-party tools (if applicable)
• Deployment: Packaging/distribution

### Out of Scope (Future Considerations) - Bullet list

---

## User Stories

Req: Yes | Format: H2 with H3 subsections

### Primary User Stories - Numbered list:

1. **As a [role], I want to [action]**, so that [benefit].
   - Example: "[concrete example]"

### Technical User Stories - Same format (optional)

---

## Core Architecture & Patterns

Req: Yes | Format: H2 with H3 subsections

### Architecture

• Architecture style/pattern name
• High-level description
• Directory structure (code block)

### Key Patterns - Numbered, bold-labeled:

**1. [Pattern]**
- [Description]
- [Code example if helpful]

---

## Tools (AI Agents) OR ## Features (Others)

Req: Yes | Format: H2 with H3 per tool/feature

For AI Agents:

### Tool N: `tool_name`

**Purpose:** [What it does]
**Operations:**
- [Operation 1]
- [Operation 2]

**Key Features:**
- [Notable feature 1]

For Other Products:

### Feature N: [Name]

**Description:** [What users get]
**Components:**
- [Component 1]

**User Experience:**
- [Interaction pattern]

---

## Technology Stack

Req: Yes | Format: H2 with H3 per category

### [Category]

- **[Tool]** (version) - [Purpose]

Categories: Backend / Frontend / Database / Infrastructure / Testing

---

## Parallel Execution Architecture

Req: Yes | Format: H2 with H3 subsections

Purpose: Enable independent team/agent workstreams to execute concurrently

## Parallel Execution Architecture

### Team Structure

**Workstream Teams:**
- **Team A - [Name]**: [Responsibility scope]
  - Focus: [What they build]
  - Expertise: [Required skills]
  - Works with: [Other teams they depend on]

- **Team B - [Name]**: [Responsibility scope]
  - Focus: [What they build]
  - Expertise: [Required skills]
  - Works with: [Other teams]

### Interface Contracts

**CRITICAL**: These contracts enable parallel development

#### Contract 1: [Name] (Team A → Team B)

**Provider:** Team A
**Consumer:** Team B

**Interface Definition:**
```[language]
# API endpoint, data schema, or contract definition

Delivery Timeline: [When available for consumption] Mock Strategy: [How Team B can work before Team A completes]

Contract 2: [Name]

[Repeat structure for each interface]

Shared Context

What ALL teams must know:

• [Global architectural decision 1]
• [Global constraint 1]
• [Naming convention 1]

Central Configuration:

```[language]

Shared config that all teams use

### Synchronization Points

**Integration Checkpoint 1:** [When]
- **Participants:** Teams A, B, C
- **Purpose:** [What gets validated]
- **Success Criteria:** [How to verify integration works]

**Integration Checkpoint 2:** [When]
- [Repeat structure]

### Dependency Graph

Phase 1 (Parallel) ├── Workstream A: [Independent] ├── Workstream B: [Independent] └── Workstream C: [Independent] ↓ Phase 2 (After Phase 1 completes) ├── Workstream D: [Depends on A, B] └── Workstream E: [Depends on C] ↓ Phase 3 (Final Integration) └── Workstream F: [Depends on D, E]

### Risk Mitigation for Parallel Work

**Risk:** [Parallel workstreams diverge or create conflicts]
**Mitigation:**
- [Strategy 1 - e.g., daily integration tests]
- [Strategy 2 - e.g., shared type definitions]

**Risk:** [Blocked waiting for dependencies]
**Mitigation:**
- [Strategy - e.g., mock implementations, contract-first development]

---

## Security & Configuration

Req: Yes | Format: H2 with H3/H4 subsections

### Authentication (if applicable)

#### **[Method]:** [Details]
[Explanation paragraph or bullets]

### Deployment Method & Data Access (if applicable)

#### **[Mechanism]:**
[Description]

#### **How It Works:**
1. [Step 1]
2. [Step 2]

#### **Security Benefits:**
- [Benefit 1]

### Configuration Management

#### **Environment Variables (`.env`):**
```[language]
# Variables

[Orchestration] Configuration:

```[language]

Config file

#### **Configuration Class:**
```[language]
# Settings code

**`### Security Scope`**
```markdown
#### **In Scope:**
- [Measure 1]

#### **Out of Scope (Keep Simple):**
- [Excluded 1]

---

## API Specification

Req: If applicable | Format: H2 with H3 sections

### External API Integrations (if applicable) - H4 per API:

#### API: [Name] v[Version]

**Documentation:** [URL]
**Research Document:** `.agents/research/[api]-research.md` (if applicable)
**Purpose:** [Why used]
**Authentication:** [Method]

**Supported Features:**
- [Feature X]: ✓ Supported via [method]
- [Feature Y]: ✗ Not supported - Alternative: [approach]

**Integration Pattern:**
```[language]
# Init, usage, error handling examples

Compatibility Notes:

• [Requirement/limitation]

Validation Strategy:

• POC: [Description]
• Expected: [Success criteria]
• Fallback: [Alternative]

**`### Internal Endpoints`** (if applicable) - H4 per endpoint:

```markdown
#### Endpoint: `[METHOD] [PATH]`

**Purpose:** [What it does]

**Request:**
```[language]
[Example]

Response:

```[language] [Example]

**Authentication:** [How]
**Error Responses:**
- `[Code]`: [Condition]

---

## Success Criteria

Req: Yes | Format: H2 with H3

### MVP Success Definition - Three bold-labeled lists:

**Functional Requirements:**
- [Must work 1]

**Quality Indicators:**
- [Quality bar 1]

**User Experience:**
- [UX goal 1]

---

## Implementation Phases

Req: Yes | Format: H2 with H3 per phase

CRITICAL FOR PARALLEL EXECUTION: Each phase must include:

• Parallelization Strategy: Which workstreams can run concurrently
• Team Assignments: Suggested agent roles for each workstream
• Dependencies: What must complete before this phase starts
• Integration Points: Where parallel work synchronizes
• Handoff Artifacts: What each workstream delivers to others

### Phase 1: [Name] ([Period])

**CRITICAL FOR EXTERNAL DEPENDENCIES:**
If project uses external APIs/services, Phase 1 MUST include:
- **External Service Setup**: Complete all account creation, API key generation, dashboard configuration
- **External Service Verification**: Test that ALL external services work as expected with our requirements
- **Fallback Planning**: Document what to do if external service setup/verification fails

**Goal:** [One sentence]

**Parallelization Strategy:**
- **Concurrent Workstreams**: [Number] independent workstreams
- **Can Start After**: External service verification passes (if applicable), or "Day 1" (if no external dependencies)

**Workstreams:**

#### Workstream A: External Service Setup & Verification (If Applicable)
**Assigned To:** integration-specialist
**Dependencies:** None - must happen FIRST
**Deliverables:**
- All external API accounts created and configured
- API keys/credentials securely stored
- Dashboard/console settings configured
- Connectivity tests passing
- Feature compatibility verified

**Setup Tasks:**
- Create accounts on [Service A], [Service B]
- Generate API keys and store in secure config
- Configure webhooks/callbacks if needed
- Set up OAuth apps if needed
- Test basic API connectivity

**Verification Tests:**
```bash
# Test each external service
# Verify required features are accessible
# Confirm rate limits and quotas are sufficient

Critical Checkpoints:

• [ ] All external services accessible with credentials
• [ ] Required API features verified as available
• [ ] No blocking compatibility issues discovered
• [ ] Setup documented for team reference

If Verification Fails:

• Document the blocker (missing feature, access denied, etc.)
• Evaluate alternatives (different API tier, alternative service)
• Update implementation plan OR escalate for requirement changes
• DO NOT P

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: giladresisi
• Source: giladresisi/ai-dev-env
• License: MIT

Install and usage instructions live in the source repository linked above.