nestjs（nestjs 效率）

NestJS

Purpose: Master NestJS coordinator that delivers production-ready features through comprehensive planning, parallel execution, and rigorous quality validation.

When to Use

• Feature implementation requiring multiple steps
• Complex refactoring across multiple files
• Database schema changes with migrations
• Multi-module updates
• Tasks requiring research + implementation + validation

Capabilities

• Plan-first approach with user approval
• Parallel execution of independent tasks
• Elimination of false positives through double-verification
• Quality gates (lint, typecheck, tests)
• Comprehensive code review

How to Use

/nestjs "Create a product management feature with CRUD operations"
/nestjs "Implement RBAC for user management module"
/nestjs "Refactor authentication to use new auth provider"

Skill Chaining Workflow

This skill automatically chains other skills in a specific sequence:

1. Implementation Phase - Uses api-* skills to generate code
2. Review Phase - Uses review skill (multiple iterations if needed)
3. Fix Phase - Uses nestjs skill again to fix issues
4. Validation Phase - Uses lint-fix, test-runner, api-nestjs-reviewer

Iterative Review Loop

[Implementation] → [Review] → Issues Found?
                     ↓              YES
                     └───────→ [Fix] → [Review Again]
                                      ↓
                                   No Issues?
                                      ↓
                                [Quality Gates]

Iteration Rules:

• Maximum 3 review-fix iterations to prevent infinite loops
• After 3rd iteration, escalate to user for manual intervention
• Each review must be narrower (focus on remaining issues only)

Process

PHASE 1: PLANNING

1. Understand requirements thoroughly
2. Research existing patterns using parallel Explore agents: Task('Explore user management patterns', subagent_type='Explore') Task('Find CQRS handler examples', subagent_type='Explore') Task('Analyze repository patterns', subagent_type='Explore') Task('Check auth guard usage', subagent_type='Explore')
3. Create detailed implementation plan
4. Verify plan against architecture rules
5. GET USER APPROVAL before proceeding (use AskUserQuestion if needed)

PHASE 2: IMPLEMENTATION

1. Execute plan using relevant skills (can run in parallel): Skill('api-feature-cqrs --name=Product --fields="name:string,price:number"') Skill('api-data-repository --name=Product --table=products') Skill('api-controller --name=Product --operations="create,update,delete"') Skill('api-dto --name=Product --fields="name:email,price:number"')
2. Monitor progress with TodoWrite
3. Handle errors and edge cases

PHASE 3: VALIDATION (Iterative Review Loop)

Execute this loop with max 3 iterations:

ITERATION 1:
├─ Skill('review') - Full comprehensive review
├─ If issues found:
│  ├─ Use Skill('lint-fix') for lint issues
│  └─ Fix confirmed issues manually or with sub-skills
├─ Skill('review') again - Verify fixes
└─ Continue to ITERATION 2 if issues remain

ITERATION 2:
├─ Skill('api-nestjs-reviewer') - NestJS-specific review
├─ Fix any remaining confirmed issues
├─ Skill('review') - Final verification
└─ Continue to ITERATION 3 if issues remain

ITERATION 3 (Last chance):
├─ Skill('review') with specific focus on remaining issues
├─ Attempt final fixes
└─ If issues still remain → ESCALATE TO USER

Review Escalation: After 3 iterations with unresolved issues, present:

• List of remaining issues
• Why they couldn't be auto-fixed
• Options for user (manual fix, accept as-is, different approach)

PHASE 4: QUALITY GATES

All gates must pass before marking complete:

1. Linting: Skill('lint-fix') then verify with pnpm nx lint api
2. Type Check: pnpm nx typecheck api
3. Unit Tests: pnpm nx test api or Skill('test-runner --type=unit')
4. Integration Tests: pnpm nx e2e api (if applicable)

Parallel Execution Rules

Use Parallel For

• Independent research (different parts of codebase)
• Non-dependent analysis (security + performance + patterns)
• Multi-file generation (commands, queries, events simultaneously)
• Validation tasks (multiple linters/checkers)

Use Sequential For

• Dependent operations (code must be written before reviewed)
• State changes (build must complete before deployment)
• Shared resources (single file being modified)

False Positive Elimination

Research Findings

Before reporting any finding:

1. File Existence Check - Verify file actually exists
2. Pattern Usage Check - Confirm pattern is actually used (not just defined)
3. Cross-Reference - Check against multiple sources
4. Code Behavior - Test against actual codebase behavior

Review Findings

Before fixing any issue:

1. Double-Verify - Use review skill again
2. Check Context - Ensure fix doesn't break existing functionality
3. Minimal Changes - Fix only what's confirmed broken
4. Re-Review - Confirm fix resolved the issue

NestJS Architecture Rules

CQRS Pattern

• Commands in commands/ directory
• Queries in queries/ directory
• Events in events/ directory
• Handlers decorated with @CommandHandler(), @QueryHandler(), @EventHandler()

Multi-Tenancy

• All tables have organization_id column
• All queries filter by tenant
• Controllers extract tenant from request context
• Cache keys are tenant-prefixed

Security

• Protected endpoints have guards (@UseGuards(JwtAuthGuard))
• PII fields encrypted at rest
• Audit logging on state changes
• Input validation on all endpoints
• Role-based permissions checked

Best Practices

• Transactions for multi-step writes
• Proper error handling with specific exceptions
• OpenAPI documentation complete
• OpenTelemetry tracing on handlers
• Cache invalidation on updates

Available Skills

Core Skills

• api-feature-cqrs - Generate complete CQRS feature
• api-data-repository - Generate repository with BaseRepository
• api-controller - Generate REST controller with OpenAPI
• api-dto - Generate DTOs with validation
• api-feature-module - Generate NestJS module

Infrastructure Skills

• api-auth-guards - Generate authentication/authorization
• api-cache-redis - Generate Redis cache service
• api-queue-bullmq - Generate BullMQ job queues
• api-event-rabbitmq - Generate RabbitMQ events
• api-tracing-otel - Generate OpenTelemetry instrumentation

Testing Skills

• api-test-unit - Generate unit tests
• api-test-integration - Generate integration tests
• api-test-e2e - Generate E2E tests
• test-runner - Run all tests

Analysis Skills

• api-nestjs-reviewer - Review NestJS code
• api-test-runner - Run and analyze tests
• api-migration-verifier - Verify database migrations
• api-openapi-reviewer - Review API documentation

Utility Skills

• lint-fix - Fix linting issues
• review - Comprehensive code review
• debugger - Diagnose and fix bugs
• refactor - Refactor code

Completion Criteria

A task is complete ONLY when:

• User requirements fully met
• All todos marked completed
• Code review passes (both skills)
• Linting passes with no errors
• Type checking passes
• All unit tests pass
• Integration tests pass (if applicable)
• No confirmed issues remain
• User is satisfied

Exit Conditions

STOP and ask user if:

• Requirements are unclear
• Multiple valid approaches exist
• User approval needed for plan
• Critical issue found that needs decision
• Unexpected blocking issue

Related Skills

• create-feature - Full-stack feature generation
• create-crud - CRUD operations using CQRS
• orchestrator - General-purpose task coordination

Related Standards

• .claude/rules/orchestrator-parallel-execution.md - Parallel execution patterns
• .claude/rules/orchestrator-execution-checklist.md - Pre/during/post checklists
• apps/api/CLAUDE.md - API-specific patterns and conventions