Expect Review (2026): Pricing, Features & Honest Verdict

TLDR

Expect is a free CLI tool that lets AI coding agents test code changes in real browsers with actual auth from your local cookies. Best for: developers using AI coding agents who need automated browser testing. Price: free, open source. Rating: 7.9/10

What is Expect?

Expect bridges the gap between AI-generated code and browser-based testing. It scans your git diff, generates an AI test plan, lets you approve it in an interactive terminal UI, then executes it in a live Playwright browser using real authentication extracted from your local browser cookies. No mock sessions. No test scripts. Just change code and let Expect verify it works.

Built by Aiden Bai at Million Software (YC W24), the same team behind Million.js (40k+ GitHub stars). The company has raised $14.1M in funding. Expect launched in March 2026 and gained 952 GitHub stars in two weeks. The tool works with Claude Code, Cursor, GitHub Copilot, and Codex CLI.

What Are Expect’s Key Features?

Diff-Aware Test Generation

Expect scans your unstaged changes or branch diffs and generates tests specifically for what changed. If you modified a login form, it creates tests for the login flow. If you changed a pricing page, it tests the pricing display. No need to maintain a separate test suite. Tests are generated on the fly based on actual code changes.

Real Auth via Cookie Extraction

Expect extracts authentication cookies from your Chrome, Firefox, or Safari profiles and injects them into the test browser. This means tests run with your actual logged-in session. No more setting up test accounts, managing credentials, or writing auth setup code. You test the app as a real user would see it.

Multi-Browser Support

Tests run on Chromium, Firefox, and WebKit through Playwright. Cross-browser testing catches issues that single-browser testing misses. The same test plan executes across all three engines without modification.

Session Recording and CI Integration

Every test run is recorded for replay. If a test fails, you watch exactly what happened in the browser. The -y flag enables headless mode with exit codes for CI pipeline integration, so Expect can run automatically on every push.

How Much Does Expect Cost?

Expect is free under the FSL-1.1-MIT license (Fair Source License). Free for most use, with restrictions on commercial use in directly competing products. The license converts to full MIT after two years. No paid tiers have been announced. Compare this to Browserbase at $50-500/mo for cloud browser infrastructure and Checkly at $30-100/mo for e2e testing. Free is hard to beat, especially with $14.1M in funding backing continued development.

Who is Expect Best For?

Use Expect if you write code with AI agents and want to verify changes work in a real browser without writing test scripts. The diff-aware approach means you never maintain a test suite. The real auth extraction means you skip the test account setup. It is especially powerful for frontend developers making rapid changes with AI assistance.

Skip Expect if you need comprehensive test coverage across your entire application (traditional e2e frameworks are better for that). Also skip it if you need cloud-native testing without local browser profiles, as Expect currently requires local cookie extraction.

Best Expect Alternatives

Playwright

Playwright is the browser automation framework that Expect is built on. Free and open source. More powerful and flexible but requires writing test scripts manually. Better for teams that want full control over their test suite.

Browserbase

Browserbase provides cloud browser infrastructure for AI agents. $50-500/mo. Better for teams that need cloud-based browser sessions without local machine dependencies. Does not generate tests automatically.

Checkly

Checkly offers monitoring-based end-to-end testing in CI pipelines. $30-100/mo. Better for production monitoring and alerting. Requires writing Playwright scripts manually.

Final Verdict: Is Expect Worth It?

Expect solves a real problem that every developer using AI coding agents faces: “did this change actually work?” The diff-aware test generation and real auth extraction are clever solutions that remove significant friction from the testing workflow. Free pricing with YC backing and $14.1M in funding means the tool is well-supported and likely to improve rapidly.

The caveats are minor. The FSL license means you cannot build a competing product on it (fair enough). Local cookie extraction requires a machine with browser profiles (not ideal for cloud-only workflows). And no pricing model means the monetization strategy is unclear. But for the immediate use case of testing AI-generated code changes in real browsers, Expect is the best tool available. Install it and try it on your next coding session.