Web Crawl and Search MCP Servers

• crawl4ai-mcp: Convert Web page to md or html
• searxng-mcp: Search Web with Searxng
• agent-skills-mcp: Agent Skills MCP
• gkzhb-mcp: Unified MCP server combining crawl4ai, searxng, and agent-skills

Common Environment Variables

All MCP servers support the following common environment variables (configured via common-mcp):

• MCP_TYPE: Transport type ( stdio , sse , http ) - defaults to "stdio"
• MCP_HOST: Host address for SSE/HTTP modes - defaults to "127.0.0.1"
• MCP_PORT: Listening port for SSE/HTTP modes - defaults to "8000" or "8001"(more details in project readme file)
• ENV: Environment mode (development or production) - defaults to "development"
• LOG_LEVEL: Logging level (DEBUG, INFO, WARNING, ERROR) - defaults to "DEBUG" in development, "INFO" in production
• MCP_AUTH: Comma-separated auth tokens for authentication (optional)
• DOTENV_FILE: Path to dotenv file to load environment variables from (optional)

Authentication

When MCP_AUTH is set, you need to include the auth token in the request header when connecting to the MCP server:

Authorization: Bearer <your-token>

Example:

curl -H "Authorization: Bearer my-secret-token" http://localhost:8000/mcp

For SSE type, connect to url http://localhost:8000/sse . For HTTP type, connect to url http://localhost:8000/mcp .

gkzhb-mcp Features

The gkzhb-mcp server provides a unified MCP server that combines multiple tools:

• crawl4ai: Web crawling and content extraction (convert pages to markdown/HTML)
• searxng: Web search capabilities using Searxng
• skills: Agent Skills integration for enhanced AI capabilities

You can selectively enable tools using the MCP_TOOL_LIST environment variable:

# Enable only crawl4ai and searxng
MCP_TOOL_LIST=crawl4ai,searxng uv run gkzhb-mcp

# Enable only skills
MCP_TOOL_LIST=skills uv run gkzhb-mcp