Globalping MCP Server implements the Model Context Protocol (MCP), enabling AI models and MCP-compatible assistants to access Globalping's global network measurement capabilities through natural language. It exposes two transports for MCP clients: a streamable HTTP endpoint at https://mcp.globalping.dev/mcp and a legacy SSE endpoint at https://mcp.globalping.dev/sse. The server supports common network tests including ping, traceroute, DNS, MTR, and HTTP, and provides tooling to list probe locations, view rate limits, retrieve past measurements, and compare results across targets. Authentication is available via OAuth or API tokens, with higher rate limits when a token is used. The design emphasizes smart context handling, offering detailed parameter descriptions so AI clients can intelligently select measurement types and options. Location targeting is flexible, accepting continent codes, country codes, city names, network names, ASN numbers, and cloud provider regions, as well as combinations of these fields. This MCP server enables seamless integration with Claude Desktop, Windsurf, Cursor, and other MCP-compatible assistants, enabling natural-language control of a global network.