Is the Playwright MCP Server Free to Use?

Is playwright mcp free? Both Playwright and its MCP server are free, open-source software with no license fee. The Playwright MCP server is published under the Apache-2.0 licence, per the CheckMCP audit of microsoft/playwright-mcp, and installs directly from its repository with no purchase step. Costs only show up around it — headless browser runs in CI, proxy or cloud services you choose to add — not in the connector itself. That distinction matters most once you compare it with paid device-testing services like BrowserStack.

Is Playwright MCP Free? What Its Open-Source License Means

The Playwright MCP server is Microsoft's own connector for browser automation, published under the Apache-2.0 licence, according to the CheckMCP audit of microsoft/playwright-mcp. That audit gave the project a 100/100 score and an A grade, and recorded 36,404 stars, 4 open issues, and a last commit 2 days before the audit — signs of a project that is both open and actively maintained. Apache-2.0 is a permissive licence: it lets anyone read, run, modify, and redistribute the code without paying a fee to Microsoft or to this directory. For a broader survey of this category, see MCP servers for browser automation.

Installing it costs nothing beyond your own machine time. The catalogue lists two equivalent commands: mcpizy install playwright through the MCPizy CLI, or npx -y @playwright/mcp run directly against the published package. Both pull the same open repository and expose the same 7 tools — including browser_navigate, browser_snapshot, browser_click, browser_type, browser_take_screenshot, browser_evaluate, and browser_fill_form. That's the core of the answer to is playwright mcp free — the code itself carries no licence fee or usage cap.

Where Could Costs Actually Show Up?

Playwright's documentation describes Playwright Test as an end-to-end test framework that bundles a test runner, assertions, isolation, and parallelization, and that supports Chromium, WebKit, and Firefox on Windows, Linux, and macOS, run locally or in CI, headless or headed, with native mobile emulation for Chrome on Android and Mobile Safari. None of that requires a paid licence to use. What it does require is compute: headless browsers spawned inside a continuous-integration pipeline consume minutes on whatever runner hosts them, and that runner is usually a paid service, not Playwright itself.

The same logic applies to anything you place around the MCP server: a proxy that routes requests through a specific region, a cloud VM that keeps a browser session alive between agent calls, or a monitoring service watching your automation. Those are third-party infrastructure choices, and their cost structure belongs to whichever provider you pick — it has nothing to do with the licence terms the CheckMCP audit recorded for microsoft/playwright-mcp.

Playwright Local vs. Playwright in a Paid CI Runner

Run locally, the Playwright MCP server drives a browser on your own machine — the same setup Playwright's documentation covers when it says browsers install locally, tests run headless or headed, and the whole flow can start from npm init playwright@latest, yarn create playwright, or pnpm create playwright. During that setup you're prompted to choose TypeScript or JavaScript, a tests folder name, whether to add a GitHub Actions workflow, and whether to install the Playwright browsers — all steps that happen on hardware you already own or already pay for outside of Playwright.

Move the same server into a CI runner and the picture changes: every headless browser launch now happens on a machine billed by the CI provider, whether that provider is GitHub Actions or another pipeline. Playwright's documentation explicitly recommends adding a GitHub Actions workflow for CI use, which is convenient, but the compute for that workflow is the CI provider's product, not something the Playwright MCP server sells or gates. For the exact commands to wire the server into an agent, see how to use Playwright MCP with Claude Code. Teams already running headless checks in CI often pair it with the Visual Regression on PRs recipe, which runs those checks on every pull request.

When Do You Need BrowserStack MCP Instead?

Playwright MCP automates browsers Playwright itself controls — Chromium, WebKit, and Firefox, per Playwright's documentation — which covers most day-to-day testing and scripted browsing an agent needs. BrowserStack MCP targets a different job: its page describes giving agents the full power of BrowserStack's Test Platform, aimed at cross-browser and device testing on BrowserStack's own infrastructure rather than a locally spawned browser. Reach for BrowserStack MCP when the requirement is testing against real devices or a wider browser and OS matrix than a local or CI-hosted Playwright run can reasonably cover.

The CheckMCP audit scored the two servers differently, and the gap is worth knowing before you pick one:

• Playwright — Audit score: 100/100 (A) — Licence: Apache-2.0 — Open issues: 4 — Last commit: 2 days ago
• BrowserStack — Audit score: 78/100 (C) — Licence: AGPL-3.0 — Open issues: 41 — Last commit: 3 days ago

BrowserStack MCP installs the same way Playwright MCP does — mcpizy install browserstack or npx -y @browserstack/mcp-server — but its repository carries an AGPL-3.0 licence rather than Apache-2.0, and the CheckMCP audit notes more open issues and a lower score. That doesn't make it unusable; it means the licence terms and maintenance signal differ, and cross-browser device coverage is what you're paying BrowserStack's platform for, separate from either connector's own licence. That workflow is covered by the Cross-Browser Testing on PRs recipe, which pairs BrowserStack MCP with GitHub.

FAQ

Is playwright mcp free to install and run locally?

Yes at the connector level: both mcpizy install playwright through the MCPizy CLI and npx -y @playwright/mcp run directly against the published package pull an Apache-2.0-licensed repository, per the CheckMCP audit of microsoft/playwright-mcp, with no purchase step anywhere in that path. Running the server locally uses your own machine to drive Chromium, WebKit, or Firefox, so beyond the hardware you already own, nothing additional is required to get the 7 exposed tools working inside an MCP-compatible agent.

Does the Playwright MCP server's page state a price?

No — the listing covers how to install the server, the 7 tools it exposes, its Apache-2.0 licence, and the CheckMCP audit score of 100/100, but it does not state a price for the connector itself. That's consistent with Apache-2.0 software generally, which grants free use, modification, and redistribution by design rather than gating access behind a paid tier. For a similar breakdown of another popular connector, see is GitHub MCP free to use.

Is BrowserStack MCP a free alternative to Playwright MCP?

Its repository carries an AGPL-3.0 licence, according to the CheckMCP audit, so the connector code itself is open just like Playwright's. What isn't free is what sits behind it: BrowserStack's Test Platform, the cross-browser and device-testing service the MCP server connects to, is a separate product from the licence on the connector code, and the audit's 41 open issues and lower 78/100 score are worth weighing before you rely on it for production device testing.