playwright-skill

Why we included this project

Most agent browser automation relies on pre-built scripts, but this skill works differently: the agent writes a Playwright program for the specific task at hand. That makes it the right choice when the automation itself is the deliverable, whether that means loops, assertions, multiple browser contexts, network interception, screenshots, video, or a script you want to keep and rerun. The skill stays out of the way in normal use, loading a short description first and pulling in the full API reference only when the task calls for it. It installs as a standard Agent Skill in Claude Code, Cursor, Copilot, Codex, and similar clients, or as a Claude Code plugin for teams that want to distribute it. If you already test with Playwright, this is a straightforward way to hand that same capability to your agents without maintaining a separate toolchain.