19 Agent Skills Test Your Web App With Playwright and Real Browsers

cover
8 Nov 2026

Nineteen open-licensed agent skills let an AI agent test a web app the way a person would: write the test first, open a real browser, click through the flows and report what broke. They come from Vercel Labs' agent-browser, Tech Leads Club, Anthropic, Matt Pocock, Jesse Vincent and lackeyjb.

All 19 are packaged as the QA & Testing plugin in Cuttlefish.MD, HackerNoon's license-checked skills marketplace. They follow the open Agent Skills standard, so they run in Claude Code, Codex CLI and 30+ other tools, and each is indexed in HackerNoon's SKILL.md Library.

The 19 QA and testing skills

Skill

What it does

Author

agent-browser

A browser automation CLI built for AI agents

Vercel Labs

agentcore

Runs agent-browser on AWS Bedrock AgentCore cloud browsers

Vercel Labs

chrome-devtools

Debugs, profiles and automates the browser through the Chrome DevTools MCP server

Tech Leads Club

core

The core agent-browser usage guide

Vercel Labs

demo

Records a demo video of a feature on a real desktop with a real cursor

Tech Leads Club

derive-client

Records browser traffic to reverse-engineer a site's internal API and generate a client

Vercel Labs

dogfood

Explores a web app systematically to find bugs and UX problems

Vercel Labs

electron

Automates Electron desktop apps such as VS Code, Slack and Figma

Vercel Labs

playwright-skill

Full browser automation with Playwright

lackeyjb

playwright-skill

Tech Leads Club's packaging of Playwright browser automation

Tech Leads Club

protected-vercel-deployments

Tests Vercel deployments behind SSO or Deployment Protection

Vercel Labs

slack

Works in Slack workspaces through browser automation

Vercel Labs

spec-driven-eval

Grades how completely an implementation meets its PRD, case by case

Tech Leads Club

tdd

Builds features and fixes bugs test-first, red-green-refactor

Matt Pocock

test-driven-development

Writes the failing test before any implementation code

Jesse Vincent

tlc-spec-driven

Plans and builds features in four phases: specify, design, tasks, execute

Tech Leads Club

vercel-sandbox

Runs agent-browser and Chrome inside Vercel Sandbox microVMs

Vercel Labs

webapp-testing

Tests local web apps with Playwright

Anthropic

webmcp-gen

Builds and validates experimental WebMCP tools for a page

Vercel Labs

Install all 19 in one command

/plugin marketplace add hackernoon/cuttlefish-agent-skills
/plugin install qa-testing@cuttlefish

For Codex CLI, Cursor or another agent that supports the standard, clone the repo and copy plugins/qa-testing/skills/ into your agent's skills folder, such as .agents/skills/. The agent-browser skills need Vercel's agent-browser CLI installed.

Each skill keeps its license and an ATTRIBUTION.md naming its author and source commit. Browse the full library at hackernoon.com/skills, and star Cuttlefish.MD on GitHub.