hovering agent and better tags

The LLM news cycle was crazy for the past few weeks. We are keeping an eye on all upgrades to set up the Agent with (a collection of) the performing ones for our purposes. But! We haven’t let that distract us from shipping new features:

The Good

‍

Functional

• We supercharged tags so you can filter and shard test execution. Right now only available via CI/CD and curl, but we’re working on the UI part. Organize your test suite and split test runs into smaller chunks which can all run in parallel! 🏷️
• There is still a world of HTTP websites out there. Now, you can submit a `http://` site as a test target when creating a new project. 🌍
• GOTO is our new addition to the interaction type family. Use the `go to URL` step in your test case when editing. 🏃
• We also added the Playwright `type characters` step to support rich text editors for fields where you type in manually.👇
• You can use the global step timeout in `environments settings` to shorten the timeout between steps if you have to run the tests faster. This option should be enjoyed with caution - it can lead to more frequent failures due to timing issues. ⏸️

Agent
The agent can hover now! If you tell it, it identifies a hover interaction and acts accordingly when discovering and generating test cases. Its 8 arms just got a bit more sensitive. 🐙

UI

We’ve brushed up several views to better align with the growing number of features and options. Ability to do complex stuff shouldn’t come with a complex UI. You will see changes in:

• environment settings & project preferences
• test case overview
• you can rearrange the width in the test case split view as you please, now

‍

The Complicated

You guys asked for assertion on highlighted text / element. The teenie weenie problem is that you can highlight a thing in many, many ways. Color the element itself, overlay the element with a transparent blurred colored div, <mark> element, change borders, outlines, text decoration, set a the background image … you name it. 

‍

It was a feat to teach the agent to spot most of those.

‍

---

‍

Allow me to wrap up this update on a personal note. I wrote a blog about the pros and cons of different ways testing tools deploy AI in the testing process. Heads up - it’s opinionated. I'd love to know your takes on that topic.