TesterArmy Launches e2e, an Open-Source AI Testing Framework for Web and Mobile

TesterArmy's new e2e framework reuses cached AI-agent steps to keep web and mobile tests from calling a model on every run. Released on October 19, 2026, the open-source TypeScript tool combines AI-agent actions with precise locators and deterministic checks so tests stay repeatable.

The framework remembers an agent's actions and replays compatible cached steps on later runs. When it can reuse those steps, e2e does not call the AI model again. Developers can combine agent-driven actions with precise locators and deterministic assertions in the same test.

A shared API supports web applications as well as mobile testing on iOS simulators and Android emulators. The e2e website also lists support for models from different providers, including ChatGPT, GitHub Copilot and SuperGrok.

Developers can choose an AI model and how to connect to it, including through gateways, a direct API key or a subscription. The framework is customizable, with support for user-defined testing agents, system prompts and execution engines.

The project's source code is publicly available. In CI mode, e2e automatically uses one retry and read-only trace caching. How useful could it be to replay cached test steps without making another call to an AI model?