SPARC Agents

16 specialized agents for different development tasks.

What is SPARC?

SPARC = Specification, Pseudocode, Architecture, Refinement, Completion

A methodology breaking complex development into focused phases, each handled by specialized agents.

Agent Overview

Agent	Focus	Primary Use
/sparc:specification	Requirements	Define what to build
/sparc:pseudocode	Logic	Design algorithms
/sparc:architecture	Structure	System design
/sparc:refinement	Improvement	Optimize code
/sparc:completion	Finalization	Testing & polish
/sparc:api	Endpoints	API design
/sparc:database	Schema	Data modeling
/sparc:security	Protection	Security review
/sparc:performance	Speed	Optimization
/sparc:testing	Quality	Test strategy
/sparc:devops	Deployment	CI/CD & infra
/sparc:monitoring	Observability	Logging & metrics
/sparc:migration	Transitions	Data/code migration
/sparc:integration	Connections	System integration
/sparc:ui	Interface	UI/UX design
/sparc:validation	Compliance	Standards check

Core Agents

/sparc:specification

Define requirements and specifications.

/sparc:specification User authentication system

Output:

• Functional requirements
• Non-functional requirements
• Constraints
• Acceptance criteria
• User stories

/sparc:pseudocode

Design algorithm structure before coding.

/sparc:pseudocode Password reset flow

Output:

• Step-by-step logic
• Decision points
• Data transformations
• Error handling paths

/sparc:architecture

Design system structure.

/sparc:architecture Microservices for e-commerce

Output:

• Component diagram
• Data flow
• Service boundaries
• Communication patterns
• Technology choices

/sparc:refinement

Improve existing implementation.

/sparc:refinement src/services/auth.ts

Focus areas:

• Code clarity
• Performance
• Error handling
• Edge cases
• Maintainability

/sparc:completion

Finalize and prepare for production.

/sparc:completion

Actions:

• Run full test suite
• Check coverage
• Verify documentation
• Review error handling
• Final security scan

Specialized Agents

/sparc:api

API design and validation.

/sparc:api User management endpoints

Output:

• Endpoint definitions
• Request/response schemas
• Authentication requirements
• Rate limiting rules
• Versioning strategy

/sparc:database

Database schema design.

/sparc:database User and order tables

Output:

• Table schemas
• Relationships
• Indexes
• Migrations
• Query patterns

/sparc:security

Security architecture review.

/sparc:security Review auth implementation

Checks:

• Authentication flows
• Authorization rules
• Input validation
• Encryption usage
• Secret management
• OWASP Top 10

/sparc:performance

Performance optimization.

/sparc:performance API response times

Focus:

• Query optimization
• Caching strategy
• Load testing
• Resource usage
• Bottleneck identification

/sparc:testing

Test strategy design.

/sparc:testing Payment processing module

Output:

• Test pyramid strategy
• Unit test patterns
• Integration test approach
• E2E scenarios
• Coverage targets

/sparc:devops

Deployment and infrastructure.

/sparc:devops Production deployment pipeline

Output:

• CI/CD pipeline
• Environment configuration
• Scaling strategy
• Rollback procedures
• Health checks

/sparc:monitoring

Observability design.

/sparc:monitoring User service monitoring

Output:

• Logging strategy
• Metrics collection
• Alerting rules
• Dashboard design
• Trace configuration

/sparc:migration

Data and code migration.

/sparc:migration Migrate from v1 to v2 API

Output:

• Migration plan
• Rollback strategy
• Data transformation
• Compatibility layer
• Verification steps

/sparc:integration

System integration design.

/sparc:integration Payment gateway integration

Output:

• Integration architecture
• API contracts
• Error handling
• Retry logic
• Circuit breaker patterns

/sparc:ui

UI/UX architecture.

/sparc:ui Dashboard component structure

Output:

• Component hierarchy
• State management
• Styling approach
• Accessibility requirements
• Responsive design

/sparc:validation

Standards and compliance check.

/sparc:validation Check GDPR compliance

Checks:

• Code standards
• Security compliance
• Accessibility (WCAG)
• Performance budgets
• Documentation completeness

Using SPARC Workflow

Full Development Cycle

# 1. Define requirements
/sparc:specification User profile feature

# 2. Design algorithm
/sparc:pseudocode Profile update logic

# 3. Design architecture
/sparc:architecture Profile service structure

# 4. Design database
/sparc:database Profile and preferences tables

# 5. Design API
/sparc:api Profile endpoints

# 6. Design tests
/sparc:testing Profile module tests

# 7. (Implement with /code:auto)

# 8. Refine implementation
/sparc:refinement src/services/profile.ts

# 9. Security review
/sparc:security Review profile service

# 10. Finalize
/sparc:completion

Selective Usage

Use individual agents as needed:

# Just need API design
/sparc:api New webhook endpoints

# Just need security review
/sparc:security Audit auth module

# Just need test strategy
/sparc:testing Coverage for payment module

Integration with claude-flow

SPARC agents can run in parallel using claude-flow:

// Parallel execution
swarm_init({ topology: 'mesh' })

agents_spawn_parallel([
  { type: 'architect', name: 'system-design' },
  { type: 'analyst', name: 'security-review' },
  { type: 'tester', name: 'test-strategy' }
])

Tips

• Start with /sparc:specification for new features
• Use /sparc:architecture before major refactoring
• Run /sparc:security before deploying new auth code
• Use /sparc:validation before major releases
• Combine agents for comprehensive coverage

Related

• SPARC Methodology Guide
• Claude-Flow Integration
• Planning Commands