What is Iris?

Iris is an AI-powered browser testing platform. Instead of writing selectors and scripts, you describe each test step in plain English:

Go to the login page
Enter {{TEST_EMAIL}} in the email field
Enter {{TEST_PASSWORD}} in the password field
Click the sign in button
Verify the dashboard heading is visible

Iris executes those steps in a real browser using Stagehand and Google Gemini, streams every frame live while the test runs, and stores the full result — step outcomes, screenshots, console output, and network events — for later inspection.

How it works

1. Author — open a live browser session in the web app, type steps in plain English, and watch them execute in real time as you write.
2. Run — trigger tests manually, from the API, or from CI. Each run executes in a fresh browser (local Playwright or Browserbase cloud).
3. Observe — follow runs live via streamed frames, then review step-by-step results, screenshots, console logs, and network activity.
4. Integrate — generate a project API key and use the iris CLI to gate your CI pipeline on test results.

Key features

• Plain-English steps — no selectors, no flaky scripts. The AI agent figures out how to perform each instruction on the page.
• Live authoring — an interactive browser session where steps run as you write them, with multi-tab support and replay.
• Workspace variables — store credentials and config once, reference them as {{variable}} in any step. Secrets are masked in the UI.
• Test prerequisites — chain tests so shared setup (like logging in) runs before the test under inspection.
• AI explore — point Iris at a page and let it propose tests for you.
• CI/CD integration — trigger runs from pipelines with the CLI and API keys; exit codes reflect pass/fail.
• Self-hostable — MIT-licensed, runs on PostgreSQL, Redis, and MinIO via Docker Compose.

Next steps