Skill Detail
Test agent-written web changes in a real browser and hand failures back for repair with Expect
Run an agent-native browser QA loop that reads recent code changes, generates a test plan, and returns concrete failures for the coding agent to fix.
Browser AutomationMulti-Framework
Browser Automation
Multi-Framework
Published
⭐ 3.4k GitHub stars
⬇ 24.5k/wk npm
INSTALL WITH ANY AGENT
npx skills add agentskillexchange/skills --skill test-agent-written-web-changes-in-a-real-browser-and-hand-failures-back-for-repair-with-expect
Works best when you want a reusable capability, not another fragile one-off prompt.
At a glance
Tools required
Node.js, Expect CLI, a supported coding agent such as Claude Code or Codex, local browser access, and a web app repository with recent changes to test
Install & setup
Follow the upstream init flow from https://www.expect.dev/, ensure a supported coding agent is installed, then run /expect in the repo after the agent changes a web app so Expect can generate and execute the browser test plan.
Author
Million
Publisher
Organization
Last updated
Apr 16, 2026
Quick brief
Use Expect when the job is post-codegen browser QA for a changed web app, not when a user just needs a generic browser automation or test framework. The invoke moment is specific: after an agent edits a web product, Expect reads the git diff, generates a test plan, runs it in a real browser with Playwright, and reports failures back so the agent can repair and re-verify. That scope boundary, agent-native browser validation and fix feedback for recent code changes, keeps this skill-shaped instead of collapsing into a plain testing product card.