The Elasticsearch MCP Server is Elastic’s official open-source server that connects AI agents to Elasticsearch data through the Model Context Protocol. Agents can explore indices and run queries in natural language instead of hand-written DSL. It has about 723 stars on GitHub and remains under maintenance in 2026. This entry covers the standalone MCP server project itself, a distinct tool from the Elasticsearch search engine and from similarly named listings.
Distributed as a Docker container, the server speaks MCP over stdio for local clients or streamable HTTP for web integrations, and works with clients such as Claude Desktop, Cursor and VS Code. It authenticates to Elasticsearch 8.x or 9.x clusters with an API key or username and password, keeping credentials in environment variables. Elastic notes the server is now deprecated in favor of the MCP endpoint built into Elastic Agent Builder (Elastic 9.2 and later), and it receives critical security updates only, but it remains a practical bridge for existing clusters. Tools exposed to agents include listing indices, reading mappings, running search and ES|QL queries, and inspecting shards.
The server is free and open source under the Apache-2.0 license. You bring your own Elasticsearch cluster, whether self-managed or Elastic Cloud, and pay only for that infrastructure.
It is for engineering and data teams that run Elasticsearch and want AI assistants to explore logs, search data or prototype queries conversationally. It is most useful today for existing deployments; new Elastic users should also look at the Agent Builder MCP endpoint that supersedes it.