Writing tests
AI explore
Point Iris at a page and let it propose tests for you.
Explore generates test candidates automatically. Give it a URL (and optional context about what matters to you), and Iris opens the page with the browser agent, examines what's there, and proposes plain-English tests you can review and save into your project.
How to use it
- In a project, open Explore and enter the URL to analyze.
- Optionally add context to focus the exploration — e.g. "focus on the checkout funnel" or "this is an admin panel; test the user management table".
- If the page requires login or other setup, pick an existing test as a prerequisite — its steps run first (with variables interpolated) so the agent explores the page in the right state.
- Review the generated tests and keep the ones you want; edit their steps like any hand-written test.
When it shines
- Bootstrapping coverage on a page that has no tests yet.
- Discovering scenarios you might not think to write — the agent enumerates the interactions it can see.
- Authenticated areas — with a login test as prerequisite, explore works behind sign-in walls.
Treat generated tests as drafts. Review each step for intent before enabling a generated test in CI — you own the assertions, the AI just proposes them.