Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
17 changes: 12 additions & 5 deletions solutions/search/mcp.md
Original file line number Diff line number Diff line change
@@ -1,14 +1,21 @@
---
navigation_title: MCP server
navigation_title: MCP servers
applies_to:
stack:
serverless:
---

# {{es}} Model Context Protocol (MCP) server
# Elastic Model Context Protocol (MCP) servers

Connect to your Elasticsearch data directly from any MCP Client (like Claude Desktop) using the Model Context Protocol (MCP).
Elastic offers two MCP server options for connecting agents to your {{es}} data. The Agent Builder MCP server is the recommended approach for {{es}} 9.2+ and Serverless deployments, offering full access to built-in and custom tools. For older {{es}} versions without Agent Builder, you can use the `mcp-elasticsearch` server which has a limited tool set.

This server enables connecting agents to your Elasticsearch data and allows you to interact with your Elasticsearch indices through natural language conversations.
## {{agent-builder}} MCP server
```{applies_to}
stack: preview 9.2
serverless: preview
```
Elastic 9.2.0+ and Serverless deployments provide an [Agent Builder MCP server endpoint](https://www.elastic.co/docs/solutions/search/agent-builder/mcp-server) that exposes all built-in and custom [tools](https://www.elastic.co/docs/solutions/search/agent-builder/tools) you can use to power agentic workflows.

Learn more in the [elastic/mcp-server-elasticsearch](https://github.com/elastic/mcp-server-elasticsearch?tab=readme-ov-file#elasticsearch-mcp-server) GitHub repository.
## {{es}} MCP server

If you're running older versions of Elasticsearch without Agent Builder, you can use [elastic/mcp-server-elasticsearch](https://github.com/elastic/mcp-server-elasticsearch?tab=readme-ov-file#elasticsearch-mcp-server). This MCP server enables connecting agents to your {{es}} data and allows you to interact with your {{es}} indices through natural language conversations, though with a more limited tool set compared to the Agent Builder MCP server.