home/roleplay/prompt-load-testing

Prompt Load Testing

GPTClaudeGemini··1,198 copies·updated 2026-07-14
prompt-load-testing.prompt
# Prompt — Load testing

> [🇷🇺 Русский](../ru/prompt-load-testing.md) · 🇬🇧 **English**

[← Index](./README.md) · [Standards](./standards.md) · [Template](./template.md)

tags: #prompt #performance #load-testing #qa #service #single-shot
**Status:** `🟢 Ready`
**Owner:** Pandamy619
**Last edit:** 2026-05-20
**Version:** v1
**Source idea:** [LogicBalls — Performance Test Script Generator](https://logicballs.com/ai-prompt/prompts/performance-test-script-generator)
**Delta from original:** added requirements for service-side metrics (not just client), warm-up, common false-positive pitfalls, and the canonical "behavior when context is missing" block.

---

## When to use

- When you need a baseline for load/stress/spike/soak testing quickly
- When the service is prepping for a release and you want to verify latency, throughput, and saturation points
- When you need a starter script for k6, Locust, or Artillery

## What to substitute

- `{{target_system}}` — string, e.g. `payments-api in stage, https://stage.payments.example.com`
- `{{test_type}}` — enum: `load | stress | spike | soak | latency_benchmark`
- `{{language_or_framework}}` — enum: `k6 | locust | artillery | jmeter | gatling`
- `{{concurrency}}` — string, e.g. `ramp 0→500 VU over 5m, hold 30m`
- `{{duration}}` — string, e.g. `45m total (5m ramp + 30m hold + 10m ramp-down)`
- `{{success_criteria}}` — list, e.g. `p95 < 300ms; error rate < 0.5%; throughput ≥ 200 RPS`
- `{{auth_model}}` — enum: `jwt | oauth2 | api_key | none`
- `{{test_data_strategy}}` — string, e.g. `pre-seeded 10k users; round-robin from pool; cleanup after run`

## Definition of done

Universal minimum — see [Definition of done](./standards.md#2-definition-of-done).

Specific criteria:

- The script includes a warm-up phase
- Specifies which metrics to watch in the service itself (CPU, memory, DB connections, GC), not just in the client
- Success criteria are quantitative and tied to percentiles, not averages
- A section on common false-positive conclusions is included (e.g., "p95 looks good but p99 saturates — the tail is real")

## Prompt

fill the variables

This prompt has 8 variables. Pro fills them into a ready-to-paste prompt for you — no manual find-and-replace.

{{target_system}{{test_type}{{language_or_framework}{{concurrency}{{duration}{{success_criteria}{{auth_model}{{test_data_strategy}
Unlock with Pro →

when to use it

Community prompt sourced from the open-source GitHub repo pandamy619/Promts (MIT). A "Prompt Load Testing" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.

tags

roleplaycommunitygeneral

source

pandamy619/Promts · MIT