proffesor-for-testing__agentic-qe/docs/optimization/QUICK_START.md

Skill Optimization Quick Start Guide

5-Minute Overview

This framework optimizes 41 Agentic QE skills for Claude 4.5, achieving 40-50% token reduction.

Instant Commands

# Validate one skill
npm run validate-skill .claude/skills/performance-testing/SKILL.md

# Validate all 41 skills
npm run validate-all-skills

# Generate detailed report
npm run optimize:report

The Optimization Formula

1. Add YAML Frontmatter (~80 tokens)

---
name: skill-name
category: specialized-testing
priority: high
tokenEstimate: 1300
agents: [qe-agent-1]
implementation_status: optimized
optimization_version: 1.0
last_optimized: 2025-12-02
dependencies: []
quick_reference_card: true
---

2. Insert Action-First Block (100-200 tokens)

<default_to_action>
When [context]:
1. ACTION: Step with outcome
2. ACTION: Step with validation
3. ACTION: Step with completion

**Quick Decision Matrix:**
- Scenario A → Action X
- Scenario B → Action Y

**Critical Success Factors:**
- Factor 1: Measurable criterion
- Factor 2: Observable outcome
</default_to_action>

3. Add Quick Reference Card (200-400 tokens)

## Quick Reference Card

### When to Use
- Trigger 1
- Trigger 2

### Key Actions
1. Action with outcome
2. Validation with criteria

### Tools Selection
| Tool | Use Case | Complexity |
|------|----------|------------|
| Tool A | Scenario X | Low |

### Success Criteria
✅ Metric 1: Target value
❌ Anti-pattern 1: What to avoid

### Agent Coordination
- `agent-1`: Primary role
- `agent-2`: Supporting role

4. Add Agent Coordination (150-250 tokens)

## Agent Coordination Hints

### Memory Namespace
Store data in: `aqe/{category}/{skill-name}/{data-type}`

### Coordination Flow
1. **Agent 1** → Primary action, store results
2. **Agent 2** → Retrieve results, analyze
3. **Agent 3** → Make decision

### Blackboard Patterns
Post to: `{category}/{event-type}` with priority `high`

5. Compress Core Content (300-500 tokens)

## Core Concepts

### Concept 1
**Definition:** Clear, 1-2 sentence explanation.

**Key Points:**
- Point 1 with measurable aspect
- Point 2 with observable outcome

**Example:**
```language
// Minimal, focused code
const result = operation();

## Token Savings Breakdown

| Section | Before | After | Savings |
|---------|--------|-------|---------|
| Introduction | 400 | 80 | **80%** |
| Core Content | 1,800 | 1,080 | **40%** |
| Examples | 300 | 210 | **30%** |
| Agent Integration | 0 | 150 | +150 |
| **Total** | **2,600** | **1,300** | **50%** |

## Validation Checklist

Run this before marking as `optimized`:

```bash
npm run validate-skill .claude/skills/my-skill/SKILL.md

Must Pass:

• YAML frontmatter complete
• <default_to_action> present
• Quick reference card included
• Agent coordination section added
• Token count ≤ 1,500
• All agents from manifest listed
• Implementation status = optimized

Common Errors & Fixes

"Missing frontmatter"

# Add YAML block at top of file
---
name: skill-name
category: specialized-testing
priority: high
tokenEstimate: 1300
agents: [qe-agent-1]
implementation_status: optimized
---

"Missing <default_to_action>"

# Add after frontmatter, before ## heading
<default_to_action>
When requested:
1. ACTION: Clear step
2. ACTION: Validation
3. ACTION: Completion
</default_to_action>

"Token limit exceeded"

Apply compression techniques:

• Convert paragraphs → bullet lists
• Use tables for comparisons
• Remove verbose explanations
• Consolidate redundant sections

Example: Before vs After

Before (2,600 tokens)

# Performance Testing

## Why Performance Testing Matters

Performance is not just about speed - it's a fundamental quality attribute that directly impacts user satisfaction, conversion rates, and business outcomes. Research has consistently shown that even small delays can have significant consequences...

### User Impact
When we talk about user impact, we're referring to how performance affects the end-user experience. Studies by Amazon have demonstrated that every 100 milliseconds of delay results in approximately 1% drop in conversion rates...

After (1,300 tokens - 50% reduction)

---
name: performance-testing
category: specialized-testing
priority: high
tokenEstimate: 1300
agents: [qe-performance-tester]
implementation_status: optimized
---

# Performance Testing

<default_to_action>
When performance testing requested:
1. IDENTIFY critical paths (checkout, search, APIs)
2. DEFINE SLOs: p95 < 200ms, throughput > 10K req/min
3. SELECT test type: Load → Stress → Spike → Endurance
4. RUN k6/JMeter with realistic scenarios
5. MONITOR: Response time, throughput, errors, resources
6. ANALYZE bottlenecks: DB, N+1, sync ops, memory, cache
7. REPORT with baselines and trends
</default_to_action>

## Quick Reference Card

**Impact:** 100ms delay = 1% conversion drop (Amazon)

### Test Types
- **Load:** Expected traffic → Before releases
- **Stress:** Beyond capacity → Find breaking point
- **Spike:** Sudden surge → Test auto-scaling
- **Endurance:** 24-72h → Memory leaks

3-Step Quick Optimization

For fastest optimization of a single skill:

Step 1: Copy Template (30 seconds)

# View optimized template
cat docs/templates/optimized-skill-template.md

Step 2: Extract & Compress (10 minutes)

• Copy essential content from original skill
• Apply compression patterns (bullets, tables)
• Add action-first directives
• Create quick reference card

Step 3: Validate & Test (5 minutes)

# Validate structure
npm run validate-skill .claude/skills/my-skill/SKILL.md

# Test with agent
Task("Test skill", "Execute with new format", "qe-agent-name")

Total Time: ~15 minutes per skill

Priority Order

Week 1: Critical (2 skills)

1. agentic-quality-engineering
2. holistic-testing-pact

Week 2: High Priority (12 skills)

3. tdd-london-chicago
4. api-testing-patterns
5. cicd-pipeline-qe-orchestrator
... (9 more)

Weeks 3-5: Medium & Low (27 skills)

Success Metrics

Current State:

• Total Skills: 41
• Optimized: 0
• Baseline Tokens: ~90,000

Target State:

• Optimized: 41 (100%)
• Final Tokens: ~49,500
• Reduction: 45% (40,500 tokens saved)

Resources

• Detailed Patterns: claude-4-5-patterns.md
• Full Template: optimized-skill-template.md
• Complete Guide: README.md

Next Steps

1. Read patterns guide (10 min): docs/optimization/claude-4-5-patterns.md
2. Review template (5 min): docs/templates/optimized-skill-template.md
3. Optimize first skill (15 min): Start with agentic-quality-engineering
4. Validate & iterate (5 min): Use validation script
5. Test with agent (2 min): Ensure functionality

Total: ~37 minutes to first optimized skill

---

Framework Version: 1.0.0 Ready to optimize all 41 skills for 40-50% token reduction