Kagami
Kagami is an MCP server designed exclusively for the Claude Code Web environment that enables browser automation using Playwright and Firefox. It features an automated setup process and manages secure external access through a JWT authentication proxy with integrated CA certificate support.
README
Kagami
Claude Code Web exclusive Playwright MCP server. Enables browser automation via JWT authentication proxy.
Overview
This project is designed exclusively for Claude Code Web environment and provides an MCP (Model Context Protocol) server for using Playwright.
Important: This project is designed to work only in Claude Code Web environment. It will not work in local environments or other environments.
Key Features:
- Auto-setup functionality (automatically installs required components on first startup)
- External access via JWT authentication proxy
- Automation using Firefox browser
- Automatic CA certificate import
- MCP protocol compliant
Architecture
Claude Code → mcp.py → playwright-mcp (Firefox) → proxy.py → JWT Auth Proxy → Internet
- mcp.py: MCP server entry point. Responsible for initial setup and launching proxy.py
- @playwright/mcp: Playwright MCP server implementation (Node.js)
- Firefox: Browser engine
- proxy.py: Local proxy server
- JWT Auth Proxy: Authentication proxy for external access
Setup
Automatic Setup (Recommended)
The following will be automatically set up on first startup:
- certutil installation
- @playwright/mcp installation
- Firefox installation
- Firefox profile creation
- CA certificate import
- Configuration file generation
Note:
- First startup may take 30 seconds or more
Usage
Launch as MCP Server
Configure in .mcp.json and Claude Code will automatically launch it:
{
"mcpServers": {
"playwright": {
"command": "python3",
"args": [
"playwright_mcp_claude_code_web/mcp.py"
]
}
}
}
File Structure
.
├── playwright_mcp_claude_code_web/
│ ├── mcp.py # MCP server (uses only standard library)
│ ├── setup_minimal.py # Minimal synchronous setup
│ ├── setup_mcp.py # Full asynchronous setup
│ └── README.md # Detailed documentation
├── .mcp.json # MCP server configuration
└── README.md # This file
Technical Details
Startup Flow
mcp.py uses a two-phase setup approach to avoid timeout issues:
Phase 1: Synchronous Setup (runs before responding)
-
Minimal Setup (
setup_minimal.py)- Install
@playwright/mcpvia npm - Create minimal Firefox configuration file
- Install
-
Fetch Tool List
- Start temporary
@playwright/mcpprocess - Fetch available tool definitions
- Store tool list for immediate response to
tools/listrequests - Terminate temporary process
- Start temporary
-
Start MCP Server
- Begin responding as MCP server
- Return stored tool list for
tools/listrequests - Return "setup in progress" error for
tools/callrequests
Phase 2: Asynchronous Setup (runs in background)
-
Full Setup (
setup_mcp.py- runs in background thread)- Install
certutil(for certificate management) - Install
proxy.pyvia uv - Install Firefox browser
- Create Firefox profile (
/home/user/firefox-profile) - Import CA certificates for TLS inspection
- Generate final configuration file
- Install
-
Start Services
- Start
proxy.py(localhost:18915) - Start
@playwright/mcpwith full configuration - Begin proxying all requests to
@playwright/mcp
- Start
Communication Flow (After Setup)
- Claude Code sends request to
mcp.pyvia MCP protocol (stdin/stdout) mcp.pyproxies request to@playwright/mcp@playwright/mcplaunches Firefox (proxy setting: localhost:18915)proxy.pyforwards requests to JWT authentication proxy- Access external sites through authenticated proxy
Security
- Import CA certificate for TLS inspection into Firefox profile
- All HTTPS traffic goes through JWT authentication proxy
- Firefox runs with dedicated profile (/home/user/firefox-profile)
Note: Claude Code Web environment performs TLS Inspection on all HTTPS communications for security auditing purposes. This project imports the necessary CA certificates to enable secure browsing within this monitored environment.
References
License
MIT
Recommended Servers
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.
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.
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.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
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.
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.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
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.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.