TestDriver

About TestDriver

TestDriver is an AI-powered end-to-end testing tool that integrates with GitHub, automatically running pull requests in a real desktop sandbox to find bugs and build regression tests. Instead of relying on selectors or DOM access, it uses an AI vision agent that reads the screen, clicks, and types to execute test flows described in plain English, then generates test code (using a Vitest-based format) automatically. Found elements are cached with a vision fingerprint so subsequent runs replay deterministically without repeated AI calls, and when the UI changes, the system re-invokes the AI to relocate elements and update the cache rather than failing outright.

It is aimed at engineering teams that need to test applications across web, desktop, and browser extension contexts, including third-party web apps without source access, Chrome extensions, native Windows and macOS applications, VS Code extensions, and content such as canvases, video players, iframes, OAuth flows, file uploads, PDFs, and LLM chatbots. Tests run in CI/CD pipelines, support assertions validated by an AI "judge" that returns a boolean with reasoning, and results can be exported to JUnit XML. The debugging interface provides video replay, network logs, request/response inspection, CPU and memory monitoring, and a step-by-step action log, alongside tracking of pass rates, flaky tests, and failure trends over time.

Pricing includes a free trial with no credit card required, a Pro plan at $24/month per user with 10 testing hours per month (overage billed at $3.60/hour) and community support, and an Enterprise plan with custom license-based billing, private AWS deployments, bring-your-own-keys, and enterprise support. Hosting is available via AWS.

Key features

• AI vision agent generates end-to-end tests from plain-English descriptions without selectors or DOM access
• Runs tests in a real desktop sandbox on Windows, covering web, desktop, Chrome extension, and VS Code extension apps
• Self-healing tests that re-locate UI elements and update the cache when the interface changes
• Runs as plain Vitest tests in CI/CD pipelines alongside existing workflows
• AI-judge assertions that validate outcomes with a boolean result and reasoning
• Debug view with video replay, network logs, CPU/memory monitoring, and step-by-step action logs

Frequently asked questions