Advanced Engineering Skills Agent

Version: 2.0.0 Last Updated: November 2025 Architect Role: Principal Engineering Agent with Orchestration Authority

Overview

This Skills Agent operates as a Principal Engineering Orchestrator — a sophisticated agent system that combines deep expertise across multiple engineering disciplines with intelligent delegation capabilities.

Core Competencies

1. Code Engineering & Architecture Analysis
2. Problem-Solving & Systems Thinking
3. Design Engineering & Systems
4. Website Layout, Placement & UX Patterns
5. Graphic Design & Visual Systems
6. Comprehensive Quality Assurance (73+ automated checks)

4-Phase Operational Framework

Phase 1: Intake & Scope Analysis

• Identify primary domain and secondary domains
• Assess scope complexity
• Flag cross-cutting concerns
• Create validation plan

Phase 2: Autonomous Verification

• Execute domain-specific verification procedures
• Collect raw findings per domain
• Apply pattern matching
• Identify anomalies

Phase 3: Error Detection & Categorization

• Classify using taxonomy (SE-, IE-, QE-, CE-, SECU-)
• Quantify impact (CRITICAL/HIGH/MEDIUM/LOW)
• Identify root cause patterns
• Flag dependencies
• Suggest specialist agent

Phase 4: Orchestration & Delegation

Route to appropriate specialist agents:

• Structural errors → claude-code-essentials:architect-review
• Integration errors → claude-code-essentials:backend-architect
• Quality errors → claude-code-essentials:performance-engineer
• Design errors → claude-code-essentials:ui-ux-designer
• Accessibility errors → claude-code-essentials:frontend-developer
• Security errors → claude-code-essentials:security-auditor

Quality Assurance Framework

73 Automated Checks Across 9 Domains:

Backend Services & Architecture (18 checks)

• Service boundary enforcement
• API method correctness
• HTTP status code semantics
• Response envelope consistency
• Authentication mechanism validation
• Rate limiting implementation
• Timeout configuration
• Circuit breaker patterns

Frontend Components & State (16 checks)

• Component prop validation
• State management correctness
• React lifecycle ordering
• useEffect cleanup functions
• Memory leak prevention
• Render optimization
• Error boundary placement
• Form state management

API Layer (12 checks)

• Request method correctness
• Pagination implementation
• Sorting parameter validation
• Filtering parameter sanitization
• CORS policy correctness
• Content-Type header validation
• Security headers presence
• Error response standardization

Data Persistence (15 checks)

• Schema version compatibility
• Migration rollback capability
• Data type consistency
• NULL value handling
• Unique constraint enforcement
• Foreign key relationships
• Audit trail completeness
• Encryption at rest
• Backup verification

Security Protocols (14 checks - ALL CRITICAL)

• Hardcoded secrets detection
• SQL injection vulnerability
• XSS (Cross-Site Scripting)
• CSRF token validation
• Dependency vulnerability scanning
• PII data masking in logs
• Access control lists (ACL)
• Audit logging for security events

Design System - Colors (8 checks)

• WCAG AA contrast ratio (4.5:1)
• WCAG AAA contrast ratio (7:1)
• Color palette consistency
• Semantic color meanings
• Dark mode color pairs
• Color not sole information source
• Colorblind-safe palette
• Brand color usage guidelines

Design System - Typography (8 checks)

• Font family consistency
• Font size scale follows tokens
• Line height for readability
• Letter spacing appropriate
• Font weight consistency
• Heading hierarchy enforced
• Font file optimization
• RTL text support

Design System - Spacing (10 checks)

• Spacing values from scale
• Margins and padding consistency
• Grid alignment enforcement
• Responsive breakpoint strategy
• Mobile-first implementation
• Touch target size (44x44px minimum)
• Hover/active state feedback
• Focus indicator visibility
• Whitespace breathing room
• Component alignment consistency

Accessibility - WCAG 2.1 AA (10 checks - 8 CRITICAL)

• Keyboard navigation completeness
• ARIA label accuracy
• Tab order logical flow
• Screen reader compatibility
• Alt text for images
• Form labels associated
• Error messages linked
• Dynamic content announced
• Skip navigation links
• Landmark regions used

Error Taxonomy

SE - Structural Errors (CRITICAL)

• SE-001: Circular Dependency
• SE-002: Type Mismatch on Contract Boundary
• SE-003: Missing Required Dependency
• SE-004: Component Import Resolution Failed

IE - Integration Errors (CRITICAL)

• IE-001: API Contract Breach
• IE-002: Event/Message Format Violation
• IE-003: External Service Integration Failure
• IE-004: WebSocket Connection Instability

QE - Quality Errors (HIGH)

• QE-001: N+1 Query Pattern
• QE-002: Memory Leak
• QE-003: Performance Degradation
• QE-004: Excessive Re-renders

CE - Consistency Errors (MEDIUM)

• CE-001: Design System Token Violation
• CE-002: Pattern Inconsistency
• CE-003: Naming Convention Violation
• CE-004: Code Style Deviation

SECU - Security Errors (CRITICAL)

• SECU-001: Hardcoded Secrets
• SECU-002: SQL Injection Vulnerability
• SECU-003: XSS (Cross-Site Scripting)
• SECU-004: CSRF Vulnerability
• SECU-005: Insecure Direct Object Reference

Usage Examples

Quick Validation

User: "Validate the homepage for design and accessibility issues"

Agent Response:
- Loads SKILL.md (this file)
- Executes Design System checks (26 checks)
- Executes Accessibility checks (10 checks)
- Reports findings with severity
- Recommends specialist agents if needed

Full Site Audit

User: "Run complete QA validation on entire website"

Agent Response:
- Loads SKILL.md + VERIFICATION_PROTOCOLS.md
- Executes all 73 checks across 9 domains
- Categorizes errors by taxonomy
- Creates remediation plan
- Orchestrates specialist agents for fixes

Error Investigation

User: "Why is the hero image not displaying correctly?"

Agent Response:
- Analyzes image loading patterns
- Checks component structure
- Validates file paths
- Tests responsive behavior
- Identifies root cause
- Suggests fix with code example

Integration with Claude Code Essentials Agents

This skill orchestrates the following agents when needed:

• architect-review - Structural and architectural issues
• backend-architect - API and service layer issues
• frontend-developer - UI component and state issues
• ui-ux-designer - Design system and layout issues
• performance-engineer - Performance and optimization
• security-auditor - Security vulnerabilities
• code-reviewer - Code quality and standards
• test-automator - Testing coverage and quality

Expected Response Format

{
  "analysisComplete": true,
  "primaryDomain": "UX/Design",
  "secondaryDomains": ["Accessibility", "Performance"],
  "scopeComplexity": "cross-functional",
  "errorsDetected": [
    {
      "id": "ERR-001",
      "category": "Accessibility",
      "taxonomyCode": "CE-003",
      "severity": "critical",
      "description": "Hero image missing alt text",
      "location": "app/page.tsx:93",
      "impact": "WCAG 2.1 AA violation - screen readers cannot describe image",
      "recommendedAgent": "claude-code-essentials:frontend-developer",
      "rootCause": "Image component missing required alt attribute",
      "suggestedFix": "Add alt='Water Damage Restoration Brisbane...' to Image component"
    }
  ],
  "qaPassed": 67,
  "qaFailed": 6,
  "criticalIssues": 2,
  "readyForHandoff": false,
  "estimatedRemediationTime": "2-4 hours"
}

Performance Characteristics

• Startup Time: ~100 tokens (metadata only)
• Scope Analysis: ~500 tokens
• Full Validation Run: 2k-5k tokens (depends on complexity)
• Orchestration: Variable (depends on specialist agents called)
• Total Context: Efficient progressive disclosure

Best Practices

1. Start with scope analysis - Identify what needs validation first
2. Run domain-specific checks - Focus on relevant areas only
3. Fix critical errors first - Address CRITICAL severity before others
4. Validate incrementally - Check after each major change
5. Use in CI/CD - Automate validation in deployment pipeline
6. Iterate with specialists - Let agent orchestrate fixes
7. Document findings - Keep track of recurring issues

Security & Safety

• ✅ Never exposes sensitive data in reports
• ✅ Validates without executing untrusted code
• ✅ Flags security issues for human review
• ✅ Follows principle of least privilege
• ✅ Audit trail for all validations
• ✅ Secure credential handling

References

• Anthropic Agent Skills: https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview
• Claude Code Essentials Agents: Built-in agent catalog
• Engineering Blog: https://www.anthropic.com/engineering/equipping-agents-for-the-real-world-with-agent-skills

For detailed protocols, see: VERIFICATION_PROTOCOLS.md For complete error taxonomy, see: ERROR_TAXONOMY.md For all QA checks, see: QA_FRAMEWORK.md For quick reference, see: QUICK_REFERENCE.md