Performance Testing (English)
中文版: See skill performance-testing.
Prompt: this directory's prompts/performance-testing_EN.md.
When to Use
- User mentions performance testing, performance-testing
- Need to execute this testing type or produce deliverables per Standard-version
- Trigger examples: "Generate/design/write performance test plan for the following"
Output Format Options
Markdown by default. For Excel / CSV / JSON, add at the end of your request; see output-formats.md.
How to Use
- Open the relevant file in this directory's
prompts/and copy the content below the dashed line. - Append your requirements and context (business flow, environment, constraints, acceptance criteria).
- If you need non-Markdown output, append the request sentence from
output-formats.mdat the end.
Code Examples
1. K6 Load Testing
Complete K6 performance testing example including load, stress, spike, and API testing.
Location: ../performance-testing/examples/k6-load-testing/
Includes:
- Load test script
- Stress test script
- Spike test script
- API performance test script
- Automated run scripts
- Detailed README documentation
Quick Start:
cd examples/k6-load-testing
chmod +x run-tests.sh
./run-tests.sh loadTest Coverage:
- Load testing (simulate normal business volume)
- Stress testing (find performance limits)
- Spike testing (sudden traffic)
- API performance testing (REST API)
- Custom metrics and thresholds
See: examples/k6-load-testing/README.md
Best Practices
Performance Test Design Principles
- Test Type Selection
- Load testing: Verify system performance under expected load - Stress testing: Find system performance limits - Spike testing: Test sudden traffic handling capability - Soak testing: Verify long-term stability
- Test Scenario Design
- Based on real user behavior - Reasonable think time - Gradually increase load - Include warm-up and cool-down phases
- Performance Metrics
- Response Time - Throughput/RPS - Error Rate - Concurrent Users - Resource Usage (CPU, Memory, Network)
- Threshold Setting
- Define based on business requirements - Use percentiles (p95, p99) - Set reasonable error rates - Monitor trend changes
Tool Selection Guide
| Tool | Use Case | Advantages |
|---|---|---|
| K6 | Modern performance testing | Scriptable, easy to use, cloud-native |
| JMeter | Traditional performance testing | Feature-rich, GUI, many plugins |
| Gatling | Scala/Java projects | High performance, beautiful reports |
| Locust | Python projects | Easy to learn, distributed |
| Artillery | Node.js projects | Simple configuration, CI/CD friendly |
Common Pitfalls
- ❌ Using unrealistic traffic models → ✅ Build scenarios from production-like behavior and workload mix
- ❌ Looking only at average latency → ✅ Track p95/p99, error rate, throughput, and saturation together
- ❌ Skipping baseline and warm-up phases → ✅ Establish baseline, warm-up, then apply staged load
- ❌ Ignoring bottleneck evidence → ✅ Correlate app metrics with CPU, memory, I/O, and downstream services
Troubleshooting
Detailed troubleshooting steps were moved to references/troubleshooting.md. Load it on demand to keep the main skill concise.
Reference Files
- prompts/performance-testing_EN.md — Performance testing Standard-version prompt
- output-formats.md — Markdown / Excel / CSV / JSON request instructions
- examples/k6-load-testing/ — Complete K6 example
- quick-start.md — 5-minute quick start guide
Related skills: api-testing-en, automation-testing-en, test-strategy-en, test-reporting-en.
Target Audience
- QA engineers and developers executing this testing domain in real projects
- Team leads who need structured, reproducible testing outputs
- AI users who need fast, format-ready deliverables for execution and reporting
Not Recommended For
- Pure production incident response without test scope/context
- Decisions requiring legal/compliance sign-off without expert review
- Requests lacking minimum inputs (scope, environment, expected behavior)
Critical Success Factors
- Provide clear scope, environment, and acceptance criteria before generation
- Validate generated outputs against real system constraints before execution
- Keep artifacts traceable (requirements -> test points -> defects -> decisions)
Output Templates and Parsing Scripts
- Template directory:
output-templates/
- template-word.md (Word-friendly structure) - template-excel.tsv (Excel paste-ready) - template-xmind.md (XMind-friendly outline) - template-json.json - template-csv.csv - template-markdown.md
- Parser scripts directory:
scripts/
- Parse (generic): parse_output_formats.py - Parse (per-format): parse_word.py, parse_excel.py, parse_xmind.py, parse_json.py, parse_csv.py, parse_markdown.py - Convert (generic): convert_output_formats.py - Convert (per-format): convert_to_word.py, convert_to_excel.py, convert_to_xmind.py, convert_to_json.py, convert_to_csv.py, convert_to_markdown.py - Batch convert: batch_convert_templates.py (outputs into artifacts/)
Examples:
python3 scripts/parse_json.py output-templates/template-json.json
python3 scripts/parse_markdown.py output-templates/template-markdown.md
python3 scripts/convert_to_json.py output-templates/template-markdown.md
python3 scripts/convert_output_formats.py output-templates/template-json.json --to csv
python3 scripts/batch_convert_templates.py --skip-same