polyflow-quality-playbook

name	polyflow-quality-playbook
description	Create PolyFlow quality playbooks, test plans, and risk audits for inventory, production, finance, purchasing, sales, auth, and deployment-sensitive flows.

PolyFlow Quality Playbook

Use this skill when you need to define what correct means for PolyFlow, expand test coverage, or audit a risky change before it reaches production.

When to Use This Skill

Use this skill when the request involves:

• building or updating tests for a PolyFlow feature
• reviewing a change for silent data loss, audit-trail gaps, incorrect stock movement, or broken accounting balance
• validating server actions, Prisma transactions, or domain state transitions
• preparing a quality checklist for release or refactor work
• deciding which modules deserve the strongest regression coverage

Workflow

1. Read the project context that defines the expected behavior.
   • Start with README.md
   • Read docs/FEATURES.md
   • Read docs/ARCHITECTURE.md
   • Read docs/ARCHITECTURAL_HEALTH_CHECK.md
   • Read docs/plans/HIGH_PRIORITY_IMPROVEMENTS_PLAN.md
   • Check package.json for available scripts and test tooling
2. Focus on the most failure-prone flows first.
   • Inventory stock movement and stock opname
   • Production execution and material consumption
   • Purchasing receipts, invoices, and approval flow
   • Sales order and maklon handling
   • Finance journals, payments, and period logic
3. Prefer real code and real data shapes.
   • Inspect the relevant src/actions/*, src/services/*, src/lib/*, and prisma/schema.prisma
   • Use the existing test style in src/**/*.test.ts and tests/**/*.test.ts
4. Turn findings into verification.
   • Write tests for the expected behavior, not implementation details
   • Call out edge cases, rollback paths, and invalid inputs
   • Identify where a new regression test should land
5. Verify with repo commands.
   • npm run lint
   • npm run test
   • npm run test:coverage
   • npm run build for UI changes
   • npx prisma generate when schema or client types change

Guidance

• Treat JSON.parse(JSON.stringify(...)) around Prisma data as a smell worth checking.
• Treat large in-memory aggregation, missing transaction boundaries, and unguarded state transitions as quality risks.
• Preserve audit trails unless the specification explicitly requires deletion.
• If a change touches stock, production, finance, or auth, write the strongest checks there first.
• Prefer one scenario per important failure mode instead of broad happy-path only coverage.

References

• docs/ARCHITECTURAL_HEALTH_CHECK.md
• docs/FEATURES.md
• docs/MAKLON_SALES_FLOW.md
• docs/SOP_MAKLON_SALES_OPERASIONAL.md
• docs/plans/HIGH_PRIORITY_IMPROVEMENTS_PLAN.md