kimtth/mcp-aoai-web-browsing
Free使用 Azure OpenAI 和 Playwright 的最小化 MCP 服务器/客户端实现。
FreeFree tier
About kimtth/mcp-aoai-web-browsing
A minimal server/client application implementation utilizing the Model Context Protocol (MCP) and Azure OpenAI. Built with FastMCP and Playwright, it enables AI-driven web browser control by converting MCP tool responses to the OpenAI function calling format via a custom bridge. Supports both in-process and external (stdio) MCP server connections, making it compatible with clients like Claude Desktop, Claude Code, and VS Code. Configuration requires Azure OpenAI endpoint, API key, deployment model, and API version. Uses 'uv' for Python dependency management.
Key Features
Minimal MCP server/client implementation using FastMCP
Web browser control via Playwright
Bridge converting MCP tool responses to OpenAI function calling format
Supports in-process and external (stdio) MCP server connections
Compatible with Claude Desktop, Claude Code, and VS Code clients
Uses 'uv' for Python dependency management
Open source and configurable with Azure OpenAI
Pros & Cons
Pros
- Open source with a clear minimal design
- Easy integration with Azure OpenAI services
- Supports multiple external clients (Claude, VS Code)
- Leverages Playwright for robust browser control
- Converts MCP responses to standard OpenAI function calls
Cons
- Requires an Azure OpenAI account and API key
- Limited to Azure OpenAI; no support for other LLM providers
- Dependency on specific tool versions (e.g., uv, FastMCP)
- Documentation focuses on development phase (December 2024), may change
Best For
AI-powered web browsing automationIntegrating MCP servers with OpenAI-compatible LLMsBuilding interactive browser-based assistants with Azure OpenAITesting web applications through LLM-driven interactions
FAQ
What is MCP?
MCP (Model Context Protocol) is an open protocol enabling secure, controlled interactions between AI applications and local or remote resources.
How do I configure Azure OpenAI?
Rename .env.template to .env and fill in values for AZURE_OPEN_AI_ENDPOINT, AZURE_OPEN_AI_API_KEY, AZURE_OPEN_AI_DEPLOYMENT_MODEL, and AZURE_OPEN_AI_API_VERSION.
Which external clients are supported?
The MCP server can be used with Claude Desktop, Claude Code, and VS Code via mcp.json configuration.
What is the role of Playwright?
Playwright provides web browser control capabilities, enabling the MCP server to navigate and interact with websites.
How does the bridge work?
The bridge converts MCP server tool responses to the OpenAI function calling format, enabling seamless integration with OpenAI-compatible LLMs.