--- id: context.load-prd version: 1.0.0 category: context experimental: true name: Load PRD description: Load and parse Product Requirements Document for backlog generation tags: - prd - requirements - context-loading model_requirements: min_context: 128000 recommended: - claude-haiku-4.5 agents: dependencies: requires: - context.use-modern-cli-tools inputs: - name: prd_file_arg description: Optional path to PRD file from --prd-file argument type: string required: false outputs: - prd_document - requirements_list tokens: avg: 3000 max: 6000 min: 1500 --- # Load PRD ## Objective Locate and load the Product Requirements Document (PRD), then extract all requirements into a structured format for backlog generation. ## CRITICAL: Output Format Requirement **You MUST output ONLY valid JSON. Do NOT output markdown, do NOT wrap in code blocks, do NOT add any text before or after the JSON.** Your response must be a single JSON object matching the schema defined in the "Output Format" section at the end of this prompt. ## Instructions ### Step 1: Locate PRD Document Check for PRD in priority order: 1. **User-provided file** (if `prd_file_arg` provided): - Read file at specified path - Validate it's a PRD - Use as primary source 2. **PRD.md** in `knowledge-base/`: - Check for `knowledge-base/PRD.md` - Use if exists 3. **Timestamped PRD** in `knowledge-base/`: - Search for `knowledge-base/PRD-*.md` files - Sort by timestamp (most recent first) - Read most recent file 4. **FUNCTIONAL.md** as fallback: - Check for `knowledge-base/FUNCTIONAL.md` - Use as fallback source 5. **Fail if not found**: - List available files in knowledge-base/ - Inform user: "No PRD found. Please run `/create-prd` first or provide --prd-file argument." - Exit with error ### Step 2: Parse PRD Structure Extract all key sections from the PRD: **Core sections**: - **Project Overview**: Name, description, goals - **Target Users & Personas**: User types, characteristics - **Success Metrics**: Measurable outcomes - **Functional Requirements**: All FR-XXX items with priority (P0/P1/P2) - **Non-Functional Requirements**: All NFR-XXX items (performance, security, etc.) - **User Stories**: All US-XXX items with acceptance criteria - **Technical Requirements**: Architecture, infrastructure, stack - **Dependencies**: External systems, services, prerequisites - **Constraints**: Technical, business, time, budget limitations - **Risks**: Identified risks and mitigation strategies - **Out of Scope**: Explicitly excluded items - **Timeline**: Milestones, deadlines, phases **Parse format**: - Extract markdown headings and structure - Parse requirement IDs (FR-001, NFR-002, US-003, etc.) - Extract priority markers (P0, P1, P2, Must Have, Should Have, Nice to Have) - Parse tables (if present) - Extract metadata (dates, authors, version) ### Step 3: Extract Requirements List Create structured list of all requirements: **For each functional requirement (FR-XXX)**: - ID (e.g., FR-001) - Title/Description - Priority (P0/P1/P2) - Associated user stories (if any) - Acceptance criteria - Dependencies (other requirements it depends on) - Domain hint (Frontend, Backend, Data, Infrastructure, Testing, Documentation) **For each non-functional requirement (NFR-XXX)**: - ID (e.g., NFR-001) - Category (Performance, Security, Scalability, Availability, etc.) - Description - Target/Threshold (e.g., "<200ms response time") - Priority (P0/P1/P2) **For each user story (US-XXX)**: - ID (e.g., US-001) - Story text ("As a [user], I want to [action], so that [benefit]") - Linked requirement(s) - Acceptance criteria - Priority ### Step 4: Extract Metadata **Project metadata**: - Project name - Project type (greenfield, brownfield, enhancement, migration) - Tech stack (if specified) - Timeline/milestones - Team size/velocity (if specified) **Requirements statistics**: - Total requirements count - P0 count, P1 count, P2 count - Functional vs non-functional split - User stories count ### Step 5: Validate PRD Completeness Quick validation for backlog generation: **Required for backlog generation**: - [ ] At least 1 P0 requirement exists - [ ] Requirements have clear descriptions - [ ] Priorities are assigned - [ ] Success metrics are defined **Warnings**: - [ ] Missing acceptance criteria (warn but proceed) - [ ] No dependencies documented (warn but proceed) - [ ] No user stories (warn but proceed) - [ ] No timeline specified (warn but proceed) **Completeness decision**: - **≥ 3 P0 requirements**: Sufficient for backlog - **1-2 P0 requirements**: Minimal, warn user - **0 P0 requirements**: Insufficient, suggest refining PRD ## Output Format ```json { "prd_document": { "source_file": "knowledge-base/PRD.md", "metadata": { "project_name": "Task Management Platform", "project_type": "brownfield", "version": "1.0", "last_updated": "2025-11-13", "tech_stack": { "frontend": ["React", "TypeScript"], "backend": ["Node.js", "Express"], "database": ["PostgreSQL"], "infrastructure": ["AWS", "Docker", "Kubernetes"] }, "timeline": { "target_launch": "Q2 2026", "milestones": [ { "name": "MVP", "date": "2026-03-31" }, { "name": "Beta", "date": "2026-05-15" } ] } }, "overview": { "description": "Real-time task management platform for remote teams", "goals": ["Improve task visibility", "Reduce status meeting overhead", "Enable async collaboration"] }, "target_users": { "primary": "Remote team leads (10-50 person teams)", "secondary": "Individual contributors", "personas": ["Team Lead", "Developer", "QA"] }, "success_metrics": [ { "metric": "User adoption", "target": "80% within 3 months" }, { "metric": "Status meetings reduction", "target": "30%" } ], "constraints": { "technical": ["Must use existing PostgreSQL database"], "business": ["Budget: $50k"], "time": ["Launch by Q2 2026"], "regulatory": ["GDPR compliant"] }, "risks": [ { "id": "RISK-001", "description": "Slack API rate limits", "impact": "high", "likelihood": "medium", "mitigation": "Implement caching and batching" } ], "out_of_scope": ["Mobile app (Phase 2)", "Advanced analytics dashboard"] }, "requirements_list": { "functional": [ { "id": "FR-001", "title": "Create and assign tasks", "description": "Users can create tasks with title, description, deadline and assign to team members", "priority": "P0", "domain_hints": ["Backend", "Frontend", "Data"], "user_stories": ["US-001", "US-002"], "acceptance_criteria": [ "Task can be created with required fields", "Task can be assigned to one or more users", "Assignees receive notifications" ], "dependencies": [], "technical_notes": "Requires tasks table, assignment logic, notification system" }, { "id": "FR-002", "title": "Real-time status updates", "description": "Task status changes are reflected in real-time across all connected clients", "priority": "P0", "domain_hints": ["Backend", "Frontend", "Infrastructure"], "user_stories": ["US-003"], "acceptance_criteria": [ "Status changes appear within 1 second", "No page refresh required", "Works for up to 100 concurrent users per team" ], "dependencies": ["FR-001"], "technical_notes": "Requires WebSocket or Server-Sent Events, pub/sub architecture" } ], "non_functional": [ { "id": "NFR-001", "category": "Performance", "description": "Page load time under 2 seconds (p95)", "priority": "P0", "target": "<2s", "measurement": "p95 response time" }, { "id": "NFR-002", "category": "Security", "description": "OAuth2 authentication with role-based access control", "priority": "P0", "target": "OAuth2 + RBAC", "compliance": ["GDPR", "SOC2"] }, { "id": "NFR-003", "category": "Scalability", "description": "Support 10,000 concurrent users", "priority": "P1", "target": "10k concurrent users", "measurement": "Concurrent active sessions" } ], "user_stories": [ { "id": "US-001", "requirement_id": "FR-001", "story": "As a team lead, I want to create new tasks with title, description, and deadline, so that I can organize work for my team", "acceptance_criteria": [ "Given I am a logged-in team lead, when I click 'New Task' and fill in title, description, deadline, then a new task is created", "Given I am creating a task, when I leave the title field empty, then I see an error 'Title is required'" ], "priority": "P0" } ] }, "statistics": { "total_requirements": 23, "functional_requirements": 18, "non_functional_requirements": 5, "by_priority": { "p0": 14, "p1": 7, "p2": 2 }, "by_domain": { "frontend": 6, "backend": 8, "data": 3, "infrastructure": 4, "testing": 0, "documentation": 2 }, "user_stories_count": 23 }, "validation": { "is_valid": true, "completeness_score": 0.95, "has_p0_requirements": true, "has_acceptance_criteria": true, "has_priorities": true, "has_success_metrics": true, "warnings": ["Some requirements lack explicit dependencies", "Testing requirements not explicitly documented"], "ready_for_backlog": true } } ``` ## Success Criteria - ✅ PRD document located and loaded - ✅ All requirements extracted with IDs and priorities - ✅ Requirements list structured and complete - ✅ Metadata parsed - ✅ Basic validation passed - ✅ Ready for requirements analysis stage ## Error Handling ### PRD Not Found **Issue**: Cannot locate PRD document **Action**: 1. List files in knowledge-base/ 2. Inform user: "No PRD found. Please run `/create-prd` first or specify `--prd-file` argument." 3. Exit with error (do not proceed) ### Malformed PRD **Issue**: PRD exists but doesn't parse correctly **Action**: 1. Attempt partial parsing 2. Extract what's available 3. Warn user about missing sections 4. Proceed if at least 1 P0 requirement found ### Insufficient Requirements **Issue**: PRD has no P0 requirements **Action**: 1. Warn user: "PRD has no P0 (critical) requirements" 2. Ask: "Should I proceed anyway or refine PRD first?" 3. Wait for user decision ## Notes - This prompt is focused on LOADING and PARSING only - No analysis or decomposition happens here (that's in next stage) - Structure output for easy consumption by `analyze-requirements` prompt - Validation is lightweight (just check if backlog generation is possible) - Deep requirements analysis happens in `onboard.analyze-requirements` ## REMINDER: Output Requirements **Output ONLY the JSON object. No markdown, no code blocks, no explanatory text. Start directly with `{` and end with `}`.**