Discover Data and Run a Query Workflow (Preview)

Use this workflow when you’re building a custom MCP client or AI application that calls Oracle Analytics MCP tools directly. This workflow guides you through identifying the appropriate data source, exploring its available tables, and executing a bounded query.

1. Find likely data sources.

   {
     "jsonrpc": "2.0",
     "id": 200,
     "method": "tools/call",
     "params": {
       "name": "oracle_analytics-find_matching_datasources",
       "arguments": {
         "nl_question": "Show revenue by sales channel"
       }
     }
   }

2. Describe the chosen subject area tables.

   {
     "jsonrpc": "2.0",
     "id": 201,
     "method": "tools/call",
     "params": {
       "name": "oracle_analytics-describe_data",
       "arguments": {
         "datamodelName": "Sales History Subject Area",
         "tableNames": ["SALES", "CHANNELS"]
       }
     }
   }

3. Execute a bounded ANSI SQL query. Use ANSI SQL for standard analytical queries. Use Logical SQL only when the query requires advanced semantic functions or aggregation overrides.

   {
     "jsonrpc": "2.0",
     "id": 202,
     "method": "tools/call",
     "params": {
       "name": "oracle_analytics-execute_oac_ansi_sql",
       "arguments": {
         "query": "SELECT \"Sales History Subject Area\".\"CHANNELS\".\"CHANNEL_DESC\" AS channel, SUM(\"Sales History Subject Area\".\"SALES\".\"AMOUNT_SOLD\") AS revenue FROM \"Sales History Subject Area\" GROUP BY \"Sales History Subject Area\".\"CHANNELS\".\"CHANNEL_DESC\" ORDER BY revenue DESC FETCH FIRST 100 ROWS ONLY",
         "maxRows": 100
       }
     }
   }