Specialized Agents

Overview

This document catalogs the specialized AI agents available in Claude Code for the Momentum Learning Management Platform. Each agent is optimized for specific tasks and equipped with relevant tools to maximize efficiency and code quality.

Agent Categories

1. General-Purpose Agents

general-purpose

Purpose: Multi-step task automation and complex research Tools: All tools available Use Cases:

  • Complex codebase searches requiring multiple iterations
  • Multi-step research and analysis tasks
  • Tasks requiring diverse tool combinations

When to Use: When you’re searching for keywords or files and aren’t confident you’ll find the right match in the first few tries.

2. Exploration & Planning Agents

Explore

Purpose: Fast codebase exploration and analysis Tools: All tools Thoroughness Levels:

  • quick: Basic searches and pattern matching
  • medium: Moderate exploration across key areas
  • very thorough: Comprehensive analysis across multiple locations

Use Cases:

  • Finding files by patterns (e.g., src/components/**/*.tsx)
  • Searching code for keywords (e.g., “API endpoints”)
  • Answering architectural questions (e.g., “how do API endpoints work?”)

Example:

Task: "Find all error handling in the authentication flow"
Agent: Explore (thoroughness: "medium")

Plan

Purpose: Software architecture and implementation planning Tools: All tools Use Cases:

  • Designing implementation strategies
  • Creating step-by-step plans
  • Identifying critical files
  • Evaluating architectural trade-offs

Example:

Task: "Plan implementation for multi-tenant course access"
Agent: Plan

3. Development Agents

develop-workflow-orchestrator

Purpose: Main coordinator for full-stack development workflow Tools: Read, Write, Bash, Glob, Task Responsibilities:

  • Automated testing coordination
  • Code review orchestration
  • Deployment pipeline management

Workflow: Infrastructure → Backend → Frontend → Testing → Review → Deploy

4. Testing & Quality Agents

test-engineer

Purpose: Comprehensive test creation and validation Tools: Read, Write, Edit, Bash, Grep, Glob, Task, Skill Capabilities:

  • Unit test generation
  • Integration test creation
  • E2E test design
  • Coverage analysis (target: >80%)
  • Quality assurance

Use Cases:

  • Proactive test generation for new features
  • Test coverage gap analysis
  • Test suite validation

Example:

// Automatically creates tests for:
- CourseService methods
- API endpoints
- React components
- Database queries

root-cause-analyzer

Purpose: Expert debugging and systematic problem-solving Tools: Read, Edit, Bash, Grep, Glob, Task, Skill Capabilities:

  • Root cause analysis (RCA)
  • Performance issue diagnosis
  • Production incident investigation
  • Minimal-impact bug fixes

Use Cases:

  • Complex bugs requiring deep investigation
  • Performance degradation analysis
  • Production incident response

Example:

Task: "Analyze why course enrollment is failing intermittently"
Agent: root-cause-analyzer

5. Security & Compliance Agents

security-auditor

Purpose: Vulnerability assessment and secure authentication Tools: Read, Edit, Bash, Grep, Glob, Task, Skill Focus Areas:

  • OWASP Top 10 compliance
  • Authentication flow security
  • Authorization validation
  • Vulnerability scanning
  • Secure coding practices

Use Cases:

  • Security reviews before production
  • Auth flow validation
  • CVE impact assessment

Example Checks:

// Automatically reviews:
- SQL injection vulnerabilities
- XSS attack vectors
- Authentication bypass risks
- Insecure direct object references
- Missing encryption

config-safety-reviewer

Purpose: Production configuration reliability and safety Tools: Read, Edit, Grep, Glob, Bash, Task, Skill Focus Areas:

  • Magic numbers and hardcoded values
  • Connection pool sizing
  • Timeout configurations
  • Rate limiting settings
  • Resource allocation

Use Cases:

  • Pre-production configuration review
  • Lambda timeout optimization
  • Database connection pool tuning
  • API Gateway throttling validation

Example Reviews:

// Flags issues like:
- Hardcoded AWS region: "us-east-1"
- Missing timeout: Lambda default 3s
- Pool size: Should be 2-5 for Aurora Serverless
- Magic number: pageSize = 100 (should be constant)

6. Architecture & Performance Agents

systems-architect

Purpose: Evidence-based design decisions and scalable patterns Tools: Read, Write, Edit, Grep, Glob, Bash, WebFetch, Task Capabilities:

  • Architectural pattern evaluation
  • Scalability analysis
  • System design validation
  • Long-term technical strategy

Use Cases:

  • Architecture Decision Records (ADRs)
  • System design reviews
  • Technology selection
  • Scalability planning

Example:

Task: "Evaluate caching strategy for course recommendations"
Agent: systems-architect
Output: ADR comparing Redis vs ElastiCache vs CloudFront

performance-tuner

Purpose: Application profiling and optimization Tools: Read, Edit, Bash, Grep, Glob, Task, Skill Focus Areas:

  • Application profiling
  • Bottleneck identification
  • Query optimization
  • Resource utilization
  • Scalability testing

Use Cases:

  • API response time optimization
  • Database query tuning
  • Frontend render performance
  • Memory leak detection

Performance Targets:

- Page Load Time: <2s
- API Response (P95): <300ms
- DB Query (P95): <50ms
- Lighthouse Score: >90

7. Documentation Agents

docs-writer

Purpose: Expert technical documentation creation Tools: Read, Write, Edit, Grep, Glob, Bash, WebFetch, Skill Capabilities:

  • API documentation
  • User guides
  • Architecture documentation
  • Code comments
  • README files

Requirements:

  • YAML front matter for GitHub Pages
  • Clear examples and usage
  • Proper navigation structure

Example Front Matter:

---
layout: default
title: "API Documentation"
parent: API Reference
nav_order: 1
---

8. Code Quality Agents

refactor-expert

Purpose: Clean architecture and technical debt reduction Tools: Read, Edit, Grep, Glob, Bash, Task, Skill Principles:

  • SOLID principles
  • DRY (Don’t Repeat Yourself)
  • Clean Code practices
  • Design patterns
  • Technical debt reduction

Use Cases:

  • Component modularization
  • Service layer extraction
  • Code duplication removal
  • Pattern implementation

Example Refactoring:

// Before: 250-line component
CourseGenerationForm.tsx (250 lines)

// After: Clean separation
CourseGenerationForm.tsx (100 lines)
  └── hooks/useCourseForm.ts (state logic)
  └── sections/BasicInfoSection.tsx (UI)
  └── sections/AudienceSection.tsx (UI)

9. Configuration & Setup Agents

statusline-setup

Purpose: Configure Claude Code status line settings Tools: Read, Edit Use Cases:

  • Customizing status line display
  • Configuring editor integration

claude-code-guide

Purpose: Claude Code and SDK documentation assistant Tools: Glob, Grep, Read, WebFetch, WebSearch Expertise:

  • Claude Code CLI features
  • Claude Agent SDK usage
  • Claude API (Anthropic API)
  • Tool use and integrations
  • Hooks, slash commands, MCP servers

When to Use: When users ask:

  • “Can Claude…?”
  • “Does Claude…?”
  • “How do I…?”

10. Workflow Orchestrators

workflow-orchestrator

Purpose: Mermaid-to-PDF conversion pipeline coordinator Tools: Read, Write, Bash, Glob, Task Responsibilities:

  • Workflow planning and execution
  • Agent coordination
  • State management
  • Error recovery
  • Quality assurance

Pipeline:

Extract → Generate → Rebuild → Convert → Organize

Agent Selection Guide

For Code Changes 

New Feature → Plan → develop-workflow-orchestrator
Bug Fix → root-cause-analyzer → test-engineer
Refactoring → refactor-expert → test-engineer

For Analysis 

Security Audit → security-auditor
Performance Issue → performance-tuner
Architecture Review → systems-architect
Configuration Review → config-safety-reviewer

For Documentation 

API Docs → docs-writer
Architecture Docs → docs-writer + systems-architect
User Guides → docs-writer

For Testing 

Unit Tests → test-engineer
Coverage Analysis → test-engineer
E2E Tests → test-engineer

Best Practices

1. Proactive Agent Use

Use agents proactively without explicit user requests when:

  • Security reviews are needed (security-auditor)
  • Tests should be generated (test-engineer)
  • Configuration changes are made (config-safety-reviewer)
  • Documentation is required (docs-writer)

2. Agent Chaining

Combine agents for comprehensive workflows:

1. develop-workflow-orchestrator (coordinate)
2. test-engineer (generate tests)
3. security-auditor (review security)
4. config-safety-reviewer (validate config)
5. docs-writer (document changes)

3. Parallel Execution

Run independent agents in parallel:

# Single message with multiple Task calls
Task(test-engineer) + Task(security-auditor) + Task(docs-writer)

4. Resume Capability

All agents support resumption:

// Initial call
Task(agent: "test-engineer", task: "Generate tests")
// Returns agent_id: "abc123"

// Resume later
Task(resume: "abc123", task: "Add edge case tests")

Common Workflows

Feature Development 

1. Plan (architecture)
2. develop-workflow-orchestrator (build)
3. test-engineer (test)
4. security-auditor (review security)
5. docs-writer (document)

Bug Fix 

1. root-cause-analyzer (diagnose)
2. test-engineer (reproduce with test)
3. [Fix code]
4. test-engineer (validate fix)
5. docs-writer (document if needed)

Production Deployment 

1. security-auditor (final security check)
2. config-safety-reviewer (validate settings)
3. performance-tuner (verify performance)
4. test-engineer (smoke tests)

Code Review 

1. refactor-expert (code quality)
2. security-auditor (security)
3. config-safety-reviewer (configuration)
4. test-engineer (test coverage)
5. docs-writer (documentation)

Tools Available to Agents

File Operations

  • Read: Read file contents
  • Write: Create/overwrite files
  • Edit: Modify existing files
  • Glob: Pattern-based file search
  • Grep: Content search

Execution

  • Bash: Run shell commands
  • Task: Invoke sub-agents

Web Access

  • WebFetch: Fetch web content
  • WebSearch: Search the web

Specialized

  • Skill: Execute Claude Code skills
  • NotebookEdit: Edit Jupyter notebooks
  • LSP: Language Server Protocol integration

Agent Performance Tips

1. Use Specific Agents 

❌ general-purpose for everything
✅ test-engineer for tests
✅ security-auditor for security
✅ docs-writer for documentation

2. Provide Clear Context 

// Good
Task(
  agent: "test-engineer",
  task: "Generate Jest tests for CourseService.getCourse with edge cases"
)

// Bad
Task(
  agent: "test-engineer",
  task: "Add tests"
)

3. Leverage Thoroughness Levels 

// Quick lookup
Explore(thoroughness: "quick", task: "Find Course model")

// Deep analysis
Explore(thoroughness: "very thorough", task: "Map all error flows")

Momentum-Specific Patterns

TypeScript Backend (Lambda) 

// test-engineer generates:
import { CourseService } from '@/shared/services/CourseService';

describe('CourseService', () => {
  it('should get course by id', async () => {
    const result = await CourseService.getCourse('123');
    expect(result).toBeDefined();
  });
});

Next.js Frontend 

// test-engineer generates:
import { render, screen } from '@testing-library/react';
import CourseCard from '@/components/CourseCard';

describe('CourseCard', () => {
  it('renders course title', () => {
    render(<CourseCard title="Test" />);
    expect(screen.getByText('Test')).toBeInTheDocument();
  });
});

Infrastructure (Terraform) 

# config-safety-reviewer validates:
resource "aws_lambda_function" "handler" {
  timeout     = 30  # ✅ Explicit timeout
  memory_size = 512 # ✅ Appropriate size

  environment {
    variables = {
      REGION = var.aws_region # ✅ Parameterized
    }
  }
}

Success Metrics

Code Quality

  • Zero ESLint errors
  • TypeScript strict mode compliance

  • 80% test coverage

  • SOLID principles adherence

Security

  • Zero critical vulnerabilities
  • OWASP Top 10 compliance
  • Secure authentication flows
  • Encrypted data at rest/transit

Performance

  • API P95 <300ms
  • Page load <2s
  • Database query P95 <50ms
  • 99.9% uptime

Documentation

  • All docs have YAML front matter
  • API endpoints documented
  • Architecture decisions recorded
  • Examples provided

Last Updated: 2025-12-24 Maintained By: Development Team Related Docs:

Back to top

Momentum LMS © 2025. Distributed under the MIT license.