MCP server + open-source CLI

Load Testing
from your AI agent

Claude Code, Cursor, Kiro, and Codex call StressMaster directly over MCP — no API key required. Prefer the terminal? The original CLI is still here, one npm install away.

terminal
|
Model Context Protocol

Plug into your AI coding agent

Load testing without leaving your editor. Claude Code, Cursor, Kiro, Codex, and any MCP-compatible agent can drive StressMaster directly — no API key, no copy-pasting commands.

One command to connect

$claude mcp add stressmaster stressmaster-mcp

Or add manually to your MCP config

~/.claude.json
{
  "mcpServers": {
    "stressmaster": {
      "command": "stressmaster-mcp"
    }
  }
}

No API key needed

Uses MCP sampling against the calling agent's own AI session. Bring-your-own-key fallback if your client doesn't support sampling.

8 tools, 5 resources

Full StressMaster surface exposed: parse, generate, execute, analyze, plus config and templates.

One-line install

`claude mcp add stressmaster stressmaster-mcp` — and you're done. No separate package, no daemon.

8 tools the agent can call

Full StressMaster surface area, exposed over MCP

run_load_test

Run the full pipeline: parse, generate, execute, analyze.

parse_command

Convert a natural-language test description into a LoadTestSpec.

generate_k6_script

Produce a K6 JavaScript script without executing it.

analyze_results

AI-powered analysis with bottleneck detection and recommendations.

get_config

Read the current StressMaster configuration.

set_config

Update a configuration value.

list_templates

List saved test templates available to the agent.

manage_template

Create, load, delete, or export reusable test templates.

5 resources the agent can read

Test history, templates, and config — surfaced as MCP resources

stressmaster://test-history
stressmaster://test-result/{id}
stressmaster://templates
stressmaster://template/{name}
stressmaster://config

Works with

Claude CodeClaude DesktopCursorKiroCodexWindsurfCline+ any MCP client
Install in 30 seconds

Built for the Modern Developer

Everything you need to test your APIs at scale, powered by AI and built for speed.

MCP Server

Use from Claude Code, Cursor, Kiro, and Codex. No API key required — runs over the agent's own session.

Natural Language

Describe tests in plain English. No scripting required.

Multi-Provider AI

Claude, OpenAI, Gemini, OpenRouter, and Amazon Q.

Dual Execution Engine

Built-in HTTP executor for simple tests, K6 integration for advanced scenarios.

Real-time Monitoring

Live progress tracking and metrics dashboard.

Smart Analysis

AI-powered recommendations and bottleneck detection.

7 Test Types

Spike, stress, endurance, volume, baseline, workflow, and batch.

OpenAPI Support

Parse OpenAPI specs and auto-generate test scenarios.

How It Works

From natural language to production-grade load tests in seconds.

Agent or CLI

MCP or Terminal

Call from Claude Code via MCP, or run the CLI directly

AI Parser

Multi-Provider

Claude, OpenAI, Gemini, OpenRouter, or the host agent

Script Generator

K6 + Built-in

Generates production-ready load test scripts

Orchestrator

Test Manager

Manages execution flow and configuration

Test Executor

Dual Engine

Built-in executor or K6 runtime with live monitoring

Results

Analysis & Reports

AI-powered insights, returned to your agent or terminal

Or run it from the terminal

Prefer the original CLI? It still works exactly how you remember — natural language in, K6-grade load test out.

stressmaster

Click the tabs above to see different test scenarios.

* Simulated output for illustration purposes. Actual CLI output may vary.

Comprehensive Performance Insights

Track every metric that matters. From response times to resource usage, get the full picture.

Response Times

Min

0

Avg

0

P50 (Median)

0

P90

0

P95

0

P99

0

Max

0
Throughput

Requests/sec

0

Data Transfer

0

Iterations

0
Error Rates

Success Rate

0

Failed Requests

0

Timeouts

0

HTTP 5xx

0
Resource Usage

VUs (Peak)

0

Test Duration

0

CPU Usage

0

Memory

0

Choose Your AI Provider

Works with all major AI providers. Pick the one that fits your needs and budget.

Provider:OpenAI
Model:GPT-3.5-turbo
Cost:$0.0015/1K tokens
Performance:Excellent
Setup:Easy
Provider:OpenAI
Model:GPT-4
Cost:$0.03/1K tokens
Performance:Best
Setup:Easy
Provider:Anthropic
Model:Claude 3 Sonnet
Cost:$0.003/1K tokens
Performance:Excellent
Setup:Easy
Provider:Google
Model:Gemini Pro
Cost:$0.0005/1K tokens
Performance:Good
Setup:Easy

All models also available through OpenRouter for unified access.

Get Started in Seconds

One install, two ways to use it: from your AI agent over MCP, or straight from the terminal.

$claude mcp add stressmaster stressmaster-mcp
terminal
$ npm install -g stressmaster
$ claude mcp add stressmaster stressmaster-mcp

Then ask your agent:

terminal
$ Load test https://httpbin.org/get with 100 users for 30 seconds

Wire it into your agent

One command and Claude Code can run real load tests for you. The CLI is still there too — pick whichever fits your flow.