PromptMeter - LLM powered load testing in plain English

Installation

npm install -g promptmeter

Quick Start

1. Set your API key:

# OpenAI
export OPENAI_API_KEY=sk-xxx

# Or Anthropic
export ANTHROPIC_API_KEY=sk-ant-xxx

2. Run a load test:

promptmeter "50 users hitting https://api.example.com for 30 seconds"

Usage

promptmeter [prompt] [options]

Options:
  -k, --api-key <key>      API key
  -p, --provider <name>    openai (default) or anthropic  
  -m, --model <model>      Model name
  -o, --output <file>      Export results to JSON
  --dry-run                Parse only, don't execute

Save API Key (optional)

promptmeter config --set-key sk-xxx
# or for Anthropic
promptmeter config --set-anthropic-key sk-ant-xxx

Example Prompts

# Simple load test
promptmeter "100 users send GET requests to https://api.example.com/users for 30 seconds"

# Conditional test
promptmeter "5 users hit https://google.com. If all return 200, start 10 users on https://youtube.com for 30s"

# Sequential steps  
promptmeter "Warm up with 2 users on /health for 5s, then ramp to 50 users on /products for 1 minute"

# Parallel test
promptmeter "Hit /api/users and /api/products simultaneously with 10 users each for 30 seconds"

# POST with body
promptmeter "POST to https://api.example.com/login with {username: 'test', password: 'test'} using 20 users"

# Using Anthropic
promptmeter -p anthropic "10 users GET https://httpbin.org/get for 10s"

What You'll See

Real-time terminal dashboard with:

• Active steps & progress
• Success/failure counts (4xx, 5xx, network errors)
• Latency (avg, P95)
• Requests per second

License

MIT