A Model Context Protocol (MCP) server designed to connect AI assistants with Grafana Loki for querying and analyzing log data. The Loki MCP Server exposes MCP-compliant semantics for log analysis, enabling AI clients to issue operations such as LogQL-based queries, keyword searches, and label discovery against Loki. The architecture shows a direct interaction between an AI assistant and Loki via the Loki MCP Server, which handles authentication and protocol translation. Features include: Query Logs using LogQL with support for range and instant queries, Search Logs with advanced filtering and pattern matching, Label Discovery to retrieve available labels and values for streams, and full MCP protocol compatibility. The server emphasizes performance through built-in caching and optimized query execution, and security via multiple authentication methods (basic auth and bearer tokens). It exposes three main MCP tools: query_logs, search_logs, and get_labels; each with parameters like query, keywords, labels, and time ranges. Installation is via pip install -e ., supports environment variable configuration for LOKI_URL, and can be run as a CLI service. Development and testing workflows are provided.