theeabrarrr/prd-sync-skill

PRD Sync Skill

This skill operationalizes a parallel-thinking approach to ensure strict alignment between the Product Requirements Document (PRD.md), the Database Architecture, and the Source Code Logic.

Workflow

Phase 1: Preparation

1. Load and read PRD.md completely.
2. Load any existing analytical documents like GAP_ANALYSIS.md to establish context.
3. Understand constraint: Do not write any new feature code. Only update documentation and analyze logic gaps.

Phase 2: Parallel Auditing

Agent A: Database-to-PRD Mapping

• Objective: Ensure all entities, relationships, constraints, and specifically Row-Level Security (RLS) policies defined in the PRD exist logically in the database layer.
• Actions:
  • View src/types/database.types.ts or query the DB directly to inspect the schema.
  • Map core tables (Users, Customers, Orders, Cylinders, Ledgers, etc.) to the Data Model Summary in PRD.
  • Check for missing columns (e.g., proof_url in handover logs, payment_method specifics).
  • Document gaps in a structured report.

Agent B: Code-to-PRD Mapping

• Objective: Ensure all workflows, business rules, and state transitions described in the PRD are covered by corresponding server actions, API routes, and components.
• Actions:
  • List files in src/app/actions/ to verify existence of business logic controllers.
  • Use grep_search to find implementations of PRD features like "Bulk Order Assignment", "Inventory Alerts", or "Order Cancellation".
  • Read through targeted files using view_file to verify the logic performs atomic checks and matches PRD requirements.
  • Document missing logic functions or deviations.

Phase 3: Synchronized Reporting

1. Combine findings from both Agent A and Agent B into a comprehensive PRD_SYNC_REPORT.md artifact.
2. Categorize gaps by severity (e.g., Blockers, Missing Features, Discrepancies).
3. If necessary, update the GAP_ANALYSIS.md or EXECUTION_PLAN.md to reflect the newly discovered gaps, ensuring nothing has slipped through the cracks.