Expect
Expect is a skill that reads an agent's code changes, writes a browser test plan, and runs it with Playwright against a real, logged-in session to find regressions before they ship.
Expect is a skill for testing an AI coding agent's changes in a real browser. It reads the current git diff, generates a test plan, and runs that plan with Playwright using the developer's own cookies rather than a fresh session, then reports back what broke so the agent can fix it.
It checks for performance issues such as long animation frames, INP and LCP, security problems including vulnerable npm dependencies and CSRF exposure, visual regressions such as broken hover states and links, and completeness gaps such as missing metadata or dead links.
Expect hooks into agents including Claude Code, Codex and Cursor, and runs on the developer's own machine. In CI mode it runs headless, skips cookie extraction, approves its plan automatically, and enforces a time limit; mobile testing is not yet supported.