Runs end-to-end browser tests on pages affected by your PR or branch changes using the agent-browser CLI. The workflow is strict about tooling: it only uses agent-browser for automation, which matters in environments like Claude Code where you might have competing browser tools available (it explicitly avoids Chrome MCP tools). You need a local dev server running and agent-browser installed. The 19.4K GitHub stars suggest the parent repo (compound-engineering-plugin) has serious adoption, though this specific skill is more narrowly focused on keeping browser testing consistent across AI coding environments. Good fit if you want deterministic E2E tests that work the same way regardless of which AI agent you're using.
npx skills add https://github.com/everyinc/compound-engineering-plugin --skill test-browser