Write, Run & Maintain testRigor Tests with Your AI Coding Agent

Heather Brooks

Open-source skills that let your agent — Claude Code, Codex, Cursor and others — author plain-English tests, run them against your app through a secure tunnel, debug from real results, and keep your cloud test suite in sync — straight from your repository.

How it Works:Your Tests as Code, Driven by Your Agent

testRigor runs UI tests written in plain English on real cloud browsers. These agent skills turn that into a tight development loop: your tests live as files in your repo, the testrigor command-line tool pushes them to a cloud test suite and runs them, and results come back as an exit code plus a JUnit report. Point a run at your deployed site — or at localhost, reached through testRigor’s secure tunnel.

Plain English, not selectors
Refer to what the user sees click "Sign in" never brittle CSS or XPath.

Tests as code
Test cases, reusable rules and variables are files you version alongside your app.

Run anywhere
The same tests run on local, staging or production — the URL is supplied at run time, never hard-coded.

The Test Language:Reads Like Instructions to a Person

// reuse a rule, generate data, verify an email
open url "/signup"
generate unique email, then enter into "Email" and save as "newUser"
enter "Str0ng-Pass!" into "Password"
click "Create account"
check that email to stored value "newUser" was received
check that page contains "Welcome"

What’s Inside:Three focused skills

Each skill is a self-contained capability, so the agent pulls in only the one relevant to the moment — authoring, running, or the full loop.

Install:Set Up in a Couple of Minutes

You’ll need a testRigor account (for an API token), Node.js 18+, and an AI coding agent — Claude Code, Codex, Cursor, or similar.

1. Install the testRigor CLI
   The skills drive this command-line tool.

   # Node.js 18+
   npm install -g testrigor-cli
   testrigor --version

2. Authenticate
   Generate a Personal Authentication Token in testRigor (your name → API Tokens), then export it so your agent and CI never hit an interactive prompt.

   export TESTRIGOR_API_KEY="your-personal-auth-token"

3. Add the skills to your agent
   Clone the repo. Claude Code auto-discovers skills copied into its skills directory. Codex, Cursor and others read the same Markdown — point the agent at the cloned repo or reference the files from your rules / AGENTS.md.

   git clone https://github.com/TestRigor/skills.git
   
   # Claude Code — personal (every project)
   mkdir -p ~/.claude/skills
   cp -r skills/testrigor-* ~/.claude/skills/
   
   # …or project-scoped:
   # cp -r skills/testrigor-* <your-project>/.claude/skills/</your-project>
   
   # Other agents — keep the repo in your project and reference
   # testrigor-dev-loop/SKILL.md from your agent's rules / AGENTS.md

Usage:Then just ask your agent

Point the CLI at a test suite, then describe what you want — your agent pulls in the right skill based on the task.

testrigor test-suite config --default < YOUR_SUITE_ID>

Write a testRigor test that logs in and checks the dashboard, then run it against http://localhost:3000.

My checkout test is failing — debug it from the JUnit report and fix it.

Seed data with a call api step, then verify the result in the UI.

# your app on :3000, then:
testrigor test-suite run --localhost --url "http://localhost:3000" \
  --test-cases-path "tests/**/*.txt" \
  --rules-path "rules/**/*.txt" \
  --explicit-mutations \
  --junit-report-save-path report.xml

Exit code 0 and failures=”0″ mean green — and your agent keeps fixing the test (or your app) and re-running until it gets there.

“We spent so much time on maintenance when using Selenium, and we spend nearly zero time with maintenance using testRigor.”

Keith Powe VP Of Engineering - IDT