How to Connect Elasticsearch to MCP for AI Agents

Elasticsearch with AI Agents

Elasticsearch's powerful search and analytics capabilities make it ideal for AI agent access. Since Elasticsearch exposes a REST API natively, you can connect it to MCP using AnythingMCP's REST connector.

Step-by-Step Guide

Step 1: Deploy AnythingMCP

git clone https://github.com/HelpCode-ai/anythingmcp.git
cd anythingmcp && docker compose up -d

Step 2: Create a REST Connector

Open the AnythingMCP dashboard at http://localhost:3000 and create a new REST connector. Set the base URL to your Elasticsearch cluster endpoint (e.g., https://your-cluster.es.amazonaws.com or http://localhost:9200).

Step 3: Configure Authentication

In the connector settings, configure API key, Basic auth, or AWS IAM credentials depending on your Elasticsearch setup.

Step 4: Configure Search Endpoints

Set up endpoints for Elasticsearch's search, aggregation, and index management APIs in the dashboard. Customize tool names and descriptions.

Step 5: Test and Connect

{
  "mcpServers": {
    "elasticsearch": {
      "url": "http://localhost:4000/mcp"
    }
  }
}

AI Agent Use Cases

• "Search for products matching 'wireless bluetooth headphones'"
• "What are the top 10 most searched terms this week?"
• "Show me error logs from the last hour"
• "Aggregate sales data by category for this month"
• "Find all documents related to customer onboarding"

Best Practices

1. Read-only access — Use an Elasticsearch role with read-only index permissions
2. Index aliases — Use aliases to abstract index names from AI agents
3. Query DSL — Pre-configure common search patterns as tool endpoints
4. Cluster health — Monitor cluster health and avoid overwhelming it with requests

Next Steps

• REST to MCP Guide — General REST patterns
• Database to MCP Guide — Native database connectors