This MCP server exposes Metoro's Kubernetes telemetry APIs to large language models, enabling natural language queries about your cluster. Built on the Model Context Protocol (MCP) and leveraging Metoro’s eBPF-based instrumentation, it makes telemetry and observability data produced by Kubernetes microservices accessible to an LLM through Claude Desktop App. When connected, you can ask questions about cluster health, resource usage, traces, and other metrics without native tooling, while remaining within your Metoro environment. The server is implemented in Go, built on the Golang MCP SDK, and runs locally as an executable. To use it, install Claude, clone the repository, build the binary (go build -o metoro-mcp-server), and configure Claude Desktop with your METORO_AUTH_TOKEN and METORO_API_URL. For hands-on testing, the repo provides a Live Demo Cluster with a public token. You can switch between authenticated access and the demo setup via claude_desktop_config.json. This project focuses on exposing external data sources to LLMs in a standardized MCP format for Kubernetes observability.