mario-playwright-mcp

mario-playwright-mcp

MCP server for browser automation with AI agents, enabling network payload capture and QA workflows.

Category
Visit Server

README

Playwright MCP Mario

version Node.js Playwright License

Browser MCP for AI agents — based on Microsoft Playwright MCP with extensions for network payload capture and QA workflows.


What is this?

An MCP server that lets LLMs control a browser through accessibility snapshots (no vision models needed). This fork adds network payload capture and sensitive redaction — better for API debugging and QA.

flowchart LR
    subgraph Card1["1- Default Playwright"]
        D1[Microsoft upstream]
    end

    subgraph Card2["2- Mario Playwright"]
        D2[network payloads · redaction]
    end

    subgraph Card3["3- Mario agent skills"]
        D3[qa-agent · QA workflows]
    end

    subgraph Runtime[runtime]
        direction LR
        subgraph BrowserCard["4- Browser"]
            Browser[Chromium]
        end
        subgraph ToolsCard["5- Tools"]
            T1[navigate]
            T2[click · type]
            T3[snapshot · screenshot]
            T4[network ++]
        end
    end

    Card1 --> Card2 --> Card3 --> Runtime
    BrowserCard --> ToolsCard
  • MCP = bridge that controls the browser and returns snapshots, screenshots, console, network.
  • Recommended: qa-agent skill for QA workflows.
  • Mario vs upstream: full payloads, URL filter, sensitive redaction — not just [POST] url => [200].

Quick start

1. Clone and install

git clone https://github.com/YOUR_ORG/mario-playwright-mcp.git
cd mario-playwright-mcp
npm install
npx playwright install chromium

2. Configure your MCP client

Cursor (in ~/.cursor/mcp.json or Settings → MCP):

{
  "mcpServers": {
    "playwright-mcp": {
      "command": "node",
      "args": [
        "/path/to/mario-playwright-mcp/packages/playwright-mcp/cli.js",
        "--output-dir",
        ".mcp-output"
      ]
    }
  }
}

Output (console logs, screenshots, etc.) goes to .mcp-output/ in the workspace. Add .mcp-output/ to .gitignore to keep the repo clean.

Generic config (VS Code, Claude Desktop, etc.) — use node + path to cli.js as above.

3. Use it

Open a project, start a chat, and use tools like browser_navigate, browser_snapshot, browser_network_requests.

QA Agent skill (recommended)

For QA workflows and structured test reports, use the qa-agent skill — an agent skill for QA engineers that uses this MCP for browser testing, post-action analysis (console + network), and evidence capture.


Tools

All tools the Playwright MCP exposes. ✨ = Mario-enhanced (better in this fork).

Core (always available)

Tool Description
browser_navigate Navigate to a URL
browser_navigate_back Go back
browser_snapshot Capture accessibility tree (better than screenshot for actions)
browser_take_screenshot Take a screenshot
browser_console_messages Get console logs
browser_network_requests List network requests — Mario: adds includePayloads, url filter, maxBodySize, sensitive redaction
browser_click Click an element
browser_type Type text into an element
browser_hover Hover over an element
browser_drag Drag and drop
browser_fill_form Fill multiple form fields
browser_select_option Select dropdown option
browser_press_key Press a key
browser_resize Set viewport size — browser opens at native resolution by default; use to test Mobile (375×667), Tablet (768×1024), Desktop (1920×1080)
browser_evaluate Run JavaScript on the page
browser_run_code Run Playwright code snippet
browser_file_upload Upload files
browser_handle_dialog Accept/dismiss dialogs
browser_wait_for Wait for text or time
browser_tabs List, create, close, or switch tabs
browser_close Close the browser
browser_install Install the browser (if missing)

Opt-in (pass --caps=…)

Capability Tools
--caps=pdf browser_pdf_save — Save page as PDF
--caps=vision browser_mouse_move_xy, browser_mouse_click_xy, browser_mouse_drag_xy, browser_mouse_down, browser_mouse_up, browser_mouse_wheel — Coordinate-based actions
--caps=testing browser_generate_locator, browser_verify_element_visible, browser_verify_list_visible, browser_verify_text_visible, browser_verify_value — Test assertions

Extensions (vs upstream)

Network payload capture

browser_network_requests supports:

Param Description
includePayloads When true, include request/response headers and bodies
url Filter by URL substring (e.g. /login, /api/users)
maxBodySize Max response body size (default 50KB)
includeStatic Include images, fonts, etc. (default false)

Example — get full details for a login call:

browser_network_requests(url: "/login", includePayloads: true)

Output:

[POST] https://api.example.com/login => [200] OK
  Request headers: { "content-type": "application/json", ... }
  Request body: {"email":"user@example.com","password":"***"}
  Response headers: { "content-type": "application/json", ... }
  Response body: {"token":"***","user":{...}}

Sensitive keys (password, token, secret, authorization, cookie, api_key) are redacted as ***.

Workflow

  1. browser_network_requests(includeStatic: false) → list URLs
  2. browser_network_requests(url: "/api/endpoint", includePayloads: true) → headers + bodies for that request

Project structure

mario-playwright-mcp/
├── packages/
│   ├── playwright-mcp/     # MCP server
│   └── extension/          # Browser extension (upstream)
├── patches/                # Network payload patch for Playwright
├── docs/                   # Extra docs
│   └── UPSTREAM.md         # Update from Microsoft upstream
└── README.md

Update from upstream

git remote add upstream https://github.com/microsoft/playwright-mcp.git
git fetch upstream
git merge upstream/main
npm install

If Playwright version changes and the patch fails, re-apply changes in node_modules/playwright/lib/mcp/browser/tools/network.js and run npx patch-package playwright. See docs/UPSTREAM.md.


Requirements

  • Node.js 18+
  • MCP client (Cursor, VS Code, Claude Desktop, Goose, etc.)

For client-specific configs (Amp, Cline, LM Studio, etc.) and CLI options, see the upstream README.


Versioning

  • main — development branch
  • 1.1 — first Mario fork release (network payload capture, .mcp-output, etc.)
  • 1.2, 1.3, … — future releases as changes land

Use semantic versioning: patch for fixes (1.1.1), minor for features (1.2.0).


License

Apache 2.0 — see LICENSE. Based on Microsoft Playwright MCP.

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured