The shape of the system
Most teams run five tools that each hold a private copy of the same facts: which environments exist, which credentials to use, which button is the checkout button. When one changes, four copies go stale. Here there is one project, one element repository and one credential vault, and every module reads from them.
Sixteen modules ship today. The ones you will meet first are the Test Recorder, the API studio, the Performance engine, the Security engine, Accessibility audits, Smart Test Selection, Visual Regression, Analytics and the Test Data Factory. Team consoles, RBAC, the tamper-evident audit log, project management and alerting integrations sit underneath them.
The seven recording files
Every recording session writes seven JSON files. They are the contract between modules, so fields are only ever added, never removed or renamed.
- session_meta.json: who recorded what, when, in which browser, plus the ID registry that links the other six files.
- xpath_details.json: every element the recorder touched, with the locator strategies it captured for each one (elem_0001, elem_0002, ...).
- network_data.json: every request and response observed during the flow (req_0001, ...), with timing.
- performance_data.json: navigation and resource timings, derived from the network data and session metadata.
- security_data.json: cookies, headers, and security observations, cross-referenced to the requests they came from.
- ui_elements.json: the catalogue of interactive elements on each visited page.
- web_test_rerun.json: the replayable step list, referencing elements and requests by ID.
Browsers and where things run
Recording and replay run in Chromium, Chrome, Firefox, WebKit or Edge. Chrome and Edge use the installed browser channel; Firefox and WebKit use bundled builds. Network throttling and timezone emulation are Chromium-family features and are not offered on Firefox or WebKit.
Cloud tenants are hosted in Mumbai, India. The intelligence features, which are smart test selection, self-healing proposals and failure analysis, run inside the platform on local models. Your code and test data are not sent to a third-party AI provider. Enterprise customers can run the whole platform on their own infrastructure with feature parity.
What is deliberately not here
- A real-device cloud. Mobile testing is in development and is not sold. If you need physical devices today, use a device-cloud vendor alongside the platform.
- A hosted agent that explores your app on its own. Agentic testing is in development and is marked as such on the home page.
- Natural-language test authoring. Also in development. The recorder is the authoring surface that ships.
Questions people ask about this
- Can I export my tests and leave?
- Yes. Recordings export as Playwright, Cypress or Selenium code; load tests export to k6, JMeter or Locust; API collections export in open formats. Leaving costs you the shared element repository, not your tests.
Source: /documentation/platform-overview, updated 2026-09-16. Answers are checked against the shipped product on that date.
Found something out of date? The people who build the module answer the contact form on the home page, and corrections to this guide are dated above. Tell us.