TraceKit

AI web testing that acts, verifies, and explains.

Solari Hiring Challenge

An autonomous AI web testing agent that navigates real websites, verifies outcomes, captures evidence, and explains what happened.

Live Demo

https://tracekit-one.vercel.app

Project Documentation

This page

GitHub

https://github.com/hopstreax/testing-agent_v0

Core idea

Give it a URL + goal. Let it test.

1. The Problem

Traditional web testing is dependable when the workflow is known and the interface is stable, but it can become expensive to create and maintain:

• Tests often require manually writing and maintaining scripts.
• UI changes can make selectors brittle.
• Failures frequently tell developers that something failed without clearly explaining what happened.
• Complex workflows across real websites can require significant setup.

TraceKit does not claim to eliminate these problems. It explores an agent-driven approach: give the system a testing goal, let it determine the next browser action from the current page, and combine that flexibility with objective verification and evidence.