expect

Why we included this project

Verifying what a coding agent changed is often the slowest, least trustworthy part of the loop, and Expect is built for that problem. It hooks into the agents you already run, Claude Code, Codex, and Cursor among them, reads the git diff, and turns it into a test plan that runs in a real browser through Playwright. Rather than driving the page with screenshots and mouse coordinates, it works directly on the DOM, which keeps it fast and lets it catch performance regressions, broken hover states, dead links, missing metadata, and security issues like CSRF or outdated dependencies. Because it reuses your real cookies and can run headless in CI on every pull request, it works as both a local sanity check and a continuous verification step. Teams shipping agent-written frontend code who would rather not maintain a separate hand-written test suite get the most value here.