e2e

Visit Website

Introduction

e2e is an end-to-end testing framework for web and mobile apps. Instead of scripting every click, you describe a goal in natural language, such as upgrading a workspace to a paid plan, and an agent drives the app until it gets there. In the same test you then check the outcome with ordinary locators and assertions, so the deterministic part of the test stays deterministic.

Its most practical idea is record-and-replay for agent steps. When an agent step is followed by an assertion that passes, e2e records the actions the agent took. The next run replays those actions without calling the model at all, and only falls back to the agent when the app has changed. Tests that contain no agent steps need no model. That keeps cost and flakiness down, which are the two usual objections to AI-driven tests.

The framework ships separate engines. @e2e-dev/web drives Chromium, Firefox and WebKit through Playwright, and @e2e-dev/mobile drives iOS simulators and Android emulators. Running init asks which engine and which model provider you want, then writes a working project. The README lists starter templates for Vite, Next.js, Astro, Angular, Expo, SwiftUI, Jetpack Compose, Kotlin Multiplatform, Flutter and .NET MAUI, each a standalone project with a passing suite.

You bring your own model access, whether that is an existing subscription or an API key. The documentation is also shipped inside the package (node_modules/e2e/docs) so that coding agents can read it offline, a small detail that matters when Claude Code, Cursor or Codex is writing the tests for you.

Be aware of maturity. The README says e2e is in active development on the way to 1.0 and that APIs and configuration can still change. The CLI also sends anonymous usage data, such as which commands and engines run and where runs fail, but not test content, app content or credentials, and you can opt out.

The source code is at e2e-inc/e2e on GitHub under the Apache-2.0 licence, with about 13,700 stars; the most recent commit was on 11 November 2026.

Launch and background: The project grew out of TesterArmy (Launch HN, 18 June 2026); the TesterArmy site now says the hosted service is winding down and points to e2e as the open-source successor.

Features

1. Natural-language agent steps: agent.act() reaches a goal described in plain English.
2. Agent assertions: agent.assert() checks a described outcome, alongside normal expect assertions.
3. Recorded replays: Verified agent steps replay with no model calls until the app changes.
4. Web engine: Chromium, Firefox and WebKit through Playwright.
5. Mobile engine: iOS simulators and Android emulators.
6. Project scaffolding: init writes a project for the engine and model provider you choose.
7. Framework templates: Starters for Vite, Next.js, Astro, Angular, Expo, SwiftUI, Jetpack Compose, KMP, Flutter and .NET MAUI.
8. Bring your own model: Use your own subscription or API key.
9. Offline docs for agents: Documentation ships inside the npm package.
10. Telemetry opt-out: Anonymous usage data can be switched off.

Pros

1. Cheap reruns: Replaying recorded steps avoids paying for model calls on every run.
2. One framework for web and mobile: The same test style covers browsers, iOS and Android.
3. Hybrid tests: Mix agent steps with precise locators where determinism matters.
4. Free to use: Open source under the Apache-2.0 licence, so the core tool costs nothing to use.
5. Actively maintained: The repository had commits as recent as 11 November 2026.
6. Community: About 13,700 GitHub stars.

Cons

1. Pre-1.0: APIs and configuration can still change.
2. Model dependency: Agent steps need a model provider and an API key or subscription.
3. Telemetry on by default: Teams with strict policies must opt out.
4. Young ecosystem: Fewer plugins and answers online than Playwright or Cypress.

Usage and Links (Example)

npx e2e init                 # choose web or mobile and a model provider
npx e2e telemetry disable    # optional: opt out of anonymous usage data

test('a member upgrades to Pro', async ({ app, agent, screen }) => {
  await agent.act('upgrade the workspace to the Pro plan');
  await agent.assert('the invoice preview shows a prorated amount');
  await expect(screen.getByRole('status')).toContainText('Pro');
});

Tips:

1. Keep agent steps for navigation that changes often, and use locators for the final checks.
2. Commit the recorded replays so CI runs without model calls.
3. Pin the version until the project reaches 1.0.

Resources:

• e2e website

Pricing

e2e is free and open source under the Apache-2.0 licence. Model usage is billed by whichever provider you connect.

Recommended for?

1. Teams already on Playwright that want AI help for brittle navigation steps.
2. Cross-platform product teams testing web, iOS and Android with one style.
3. Coding-agent workflows where an agent writes and maintains the tests.

Conclusion

e2e is one of the more grounded takes on agent-driven testing: it pairs agent steps with real assertions and replays verified steps without the model, which addresses cost and flakiness directly.

Its pre-1.0 status is the main reason to pilot it on a slice of your suite before committing.

On CheckOps, e2e is filed under the End-to-end & functional and Release & CI gates stages. Related tools listed here include Playwright, Midscene.js, Stagehand, Maestro, Shortest and TestZeus Hercules. When comparing them, weigh the licence, the languages and platforms each supports, whether it is self-hosted or managed, and how cleanly it slots into your existing CI pipeline. Details on this page were checked against the official website, documentation and repository on 11 November 2026.