Connect Your AI Agent¶

The final step: connect your favorite AI assistant to Moltler and start querying your Elasticsearch data with natural language.

Supported Agents¶

Cursor IDE¶

Step 1: Create or edit .cursor/mcp.json in your project:

{
  "mcpServers": {
    "moltler": {
      "url": "http://localhost:9200/_escript/mcp",
      "headers": {
        "Authorization": "Basic ZWxhc3RpYy1hZG1pbjplbGFzdGljLXBhc3N3b3Jk"
      }
    }
  }
}

Step 2: Restart Cursor

Step 3: Start asking questions:

"Find all ERROR logs from the last hour"

"Show me the slowest API transactions"

"Hunt for IP 192.168.1.100 in security events"

"What's the error rate trend for the payment service?"

Claude Desktop¶

Step 1: Find your config file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Step 2: Add the Moltler server:

{
  "mcpServers": {
    "moltler": {
      "command": "curl",
      "args": [
        "-X", "POST",
        "-H", "Content-Type: application/json",
        "-H", "Authorization: Basic ZWxhc3RpYy1hZG1pbjplbGFzdGljLXBhc3N3b3Jk",
        "http://localhost:9200/_escript/mcp"
      ]
    }
  }
}

Step 3: Restart Claude Desktop

Step 4: Ask Claude to help you:

"I need to investigate a production incident. Can you find recent errors and correlate them with slow transactions?"

Cline (VS Code Extension)¶

Step 1: Open VS Code Settings (JSON)

Step 2: Add MCP configuration:

{
  "cline.mcpServers": {
    "moltler": {
      "url": "http://localhost:9200/_escript/mcp",
      "headers": {
        "Authorization": "Basic ZWxhc3RpYy1hZG1pbjplbGFzdGljLXBhc3N3b3Jk"
      }
    }
  }
}

Step 3: Restart VS Code

For Production Clusters¶

Replace the default credentials with your own:

# Generate base64 credentials
echo -n "your-user:your-password" | base64
# Output: eW91ci11c2VyOnlvdXItcGFzc3dvcmQ=

Then use in your config:

{
  "headers": {
    "Authorization": "Basic eW91ci11c2VyOnlvdXItcGFzc3dvcmQ="
  }
}

For HTTPS clusters:

{
  "url": "https://your-cluster:9200/_escript/mcp"
}

What Can You Ask?¶

Once connected, your AI agent has access to all installed skills. Here are examples by solution:

Observability¶

Security¶

Search¶

The Magic Moment¶

After connecting, try this conversation:

You: "I got paged for a production incident. The payment service is having issues. Can you help me investigate?"

Agent: Uses get_recent_errors to find errors in payment service

"I found 47 errors in the payment service in the last hour. The most common error is 'Connection timeout to payment gateway'. Let me check if there are any slow transactions..."

Uses get_slow_transactions to find latency issues

"I see the payment gateway calls are averaging 12 seconds, which is 10x higher than normal. This correlates with the timeout errors. The issue started around 9:45 AM..."

This is the power of Moltler: Your AI assistant can investigate your Elasticsearch data just like a senior engineer would.

Troubleshooting¶

Agent can't connect¶

# Test the MCP endpoint directly
curl -u elastic:password http://localhost:9200/_escript/mcp \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc": "2.0", "method": "tools/list", "id": 1}'

Skills not showing up¶

# Verify skills are installed
curl -u elastic:password http://localhost:9200/_escript \
  -H "Content-Type: application/json" \
  -d '{"query": "SHOW SKILLS"}'

Authentication errors¶

• Verify your base64 credentials are correct
• Check if your ES cluster requires API keys instead of basic auth

Next Steps¶