Playwright
Playwright is Microsoft's library for driving Chromium, Firefox, and WebKit with one API. Rechecked 2026-08-17: playwright.dev still leads with that triad. The current docs release is v1.58. Besides test runner, codegen, and trace viewer, that version ships Playwright Agents (planner, generator, healer) and an MCP server so agents can call Playwright as a tool. The repo microsoft/playwright is Apache-2.0.
This page used to read like a generic 2020 testing brochure. The product is now also an agent browser backend. Compare Cloudflare Computer or E2B if you need a hosted desktop instead of a local browser binary.
Key Features
- Three engines: Chromium, Firefox, WebKit. Headless or headed.
- Auto-wait and tracing: Built-in waits, traces, screenshots, videos.
- Codegen: Record actions into tests.
- Agents in v1.58: Planner, generator, and healer loops on top of the test runner.
- MCP: Docs now describe a Playwright MCP server for agent tool use.
Use Cases
- Cross-browser regression in CI.
- Agent products that need a real browser, not a toy HTML sandbox.
- Debugging flaky UI with the trace viewer.
Limitation: Playwright is a library you run. It is not a hosted computer. License is Apache-2.0, not a paid seat. We did not invent a price.
Pricing
| Plan | Price | Notes |
|---|---|---|
| Open source | $0 | Apache-2.0. You pay CI machines and browsers. |
Microsoft also sells Azure Playwright-style hosted runners. Recheck Azure if you need that SKU. Do not treat it as required.
Getting Started
- Read playwright.dev.
npm init playwright@latest(or the Python / Java / .NET equivalent on that page).- Run the sample test, then open the trace on failure.
- If an agent needs the browser, read the v1.58 agents / MCP docs before wrapping
page.gotoyourself.
Frequently Asked Questions
Is Playwright only a test framework?
No. Tests are the original product. v1.58 also documents agents and MCP.
Does it replace Cypress?
Only if you want first-party WebKit plus Firefox from one API. Recheck Cypress separately.
Is there a SaaS bill?
The library is free. Hosted runners, if you use them, are a different Microsoft bill.
Alternatives
- E2B: Hosted microVM when you need a disposable machine, not just a browser.
- Cloudflare Computer: Edge-hosted computer use.
- Claude Code: If the job is coding, not clicking a page.
Tips
- Start from
npm init playwright@latest, not a copied 2020 snippet. - Keep traces on in CI. That is the debug product.
- Recheck the release notes. Agents and MCP were not in the old catalog copy.
Conclusion
Playwright is still Microsoft's three-engine browser automation library, now with v1.58 agents and MCP. Start at playwright.dev, scaffold a project, then decide whether you also need a hosted sandbox like E2B.
Comments
No comments yet. Be the first to comment!
Related Tools
Agent Browser
github.com/vercel-labs/agent-browser
Vercel browser CLI for agents. Rechecked: Apache-2.0, 40,767 stars. Native Rust binary; not a Node fallback census.
Playwright Automation
github.com/lackeyjb/playwright-skill
Browser automation and testing with Playwright for web application testing, UI automation, and end-to-end testing workflows.
WebApp Testing Skill
github.com/anthropics/skills/tree/main/skills/webapp-testing
Anthropic's official toolkit for interacting with and testing local web applications using Playwright for frontend functionality verification and debugging.
Related Insights
Seven AI Coding CLIs, Six Months: No Matter How Strong the Model, Work Needs Supervision
Claude Code, Codex, opencode, pi, omp and DeepSeek Harness all have personalities. After six months of deep use I run a division of labor: pi for the fastest cheapest reviews, omp for complex PRs, DeepSeek Harness on V4 Flash for high-frequency low-cost review, and Claude Code, Qoder and Cursor for writing. No matter how strong the model, work needs supervision — ideally from an independent third party.
The Twilight of Low-Code Platforms: Why Claude Agent SDK Will Make Dify History
A deep dive from first principles of large language models on why Claude Agent SDK will replace Dify. Exploring why describing processes in natural language is more aligned with human primitive behavior patterns, and why this is the inevitable choice in the AI era.
Skills + Hooks + Plugins: How Anthropic Redefined AI Coding Tool Extensibility
An in-depth analysis of Claude Code's trinity architecture of Skills, Hooks, and Plugins. Explore why this design is more advanced than GitHub Copilot and Cursor, and how it redefines AI coding tool extensibility through open standards.