testing-load-designer_skill
- Python
5
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill williamzujkowski/cognitive-toolworks --skill testing-load-designer- SKILL.md16.9 KB
Overview
This skill designs repeatable load testing scenarios for k6, JMeter, Gatling, or Locust, including ramp-up patterns, think time modeling, and SLI validation. It produces executable test scripts, a concrete test configuration, assertion thresholds, and an execution plan tailored to the chosen tool and test type.
How this skill works
Provide a target service URL, desired test type (load, stress, spike, soak), SLI requirements, and optional tool choice. The skill validates inputs, selects sensible defaults (k6 if unspecified), maps test type to a ramp pattern, models realistic think time, and emits a tool-specific script plus JSON configs and an execution checklist. It also flags abort conditions and suggests distributed or observability integrations when relevant.
When to use it
- Validate performance before a production release or major traffic event
- Establish baselines and capacity planning for backend services
- Simulate peak, spike, stress, or extended soak conditions
- Verify SLI/SLO compliance for latency, throughput, and error rates
- Exercise distributed system resilience under sustained load
Best practices
- Verify target URL accessibility and avoid running against production without approval
- Model think time with randomized distributions to mimic real users
- Start with a smoke test and low VU counts, then scale via staged ramps
- Map SLIs to tool-native assertions (k6 thresholds, JMeter assertions, Gatling assertions, Locust custom checks)
- Use synthetic data and never include hardcoded credentials or production PII
- Integrate metrics (Prometheus/Grafana or CloudWatch/Datadog) and set clear abort criteria
Example use cases
- k6 scripted load test for e-commerce checkout with gradual ramp-up and p95 latency threshold
- JMeter stress run to find breaking point using multi-stage ramps and duration assertions
- Gatling capacity test for JVM-based service with percentile assertions and CSV parameterization
- Locust soak test to detect memory leaks over multi-hour runs with master-worker distribution
- Spike test for flash-sale simulation: instant jump to peak VUs, short sustain, and rapid drop
FAQ
k6 is recommended for modern APIs due to its JavaScript scripting, built-in thresholds, and easy CI integration. Choose Gatling for JVM-heavy environments, JMeter for complex protocol coverage, and Locust if you prefer Python.
How do I translate SLIs into VUs?
Use the VU formula: VUs = (target_RPS × response_time_seconds) / (1 - think_time_ratio). Validate assumptions with a small pilot run and adjust think time and response-time estimates.
