Overview of Developing with Oracle Analytics Cloud MCP Server (Preview)
The Oracle Analytics Cloud MCP Server provides a standardized interface for an AI agents or LLM to interact with a Oracle Analytics Cloud instance. It implements the Model Context Protocol specification, enabling natural language-to-SQL translation, and business intelligence querying.
Note:
It's critical that you understand the potential security risks before you use the Oracle Analytics Cloud MCP server. To learn about the risks, see Access the Oracle Analytics Cloud MCP Server (Preview).
Capabilities
The Oracle Analytics Cloud MCP Server provides you with these capabilities:
- Data Discovery - Enumerates available subject areas and datasets. Uses the oracle_analytics-discover_data tool. See Use the oracle_analytics-discover_data MCP Server Tool (Preview).
- Schema Introspection - Retrieves detailed column metadata including data types, aggregation rules, and relationships. Uses the oracle_analytics-describe_data tool. See Use the oracle_analytics-describe_data MCP Server Tool (Preview).
- Query Execution - Executes Oracle Logical SQL (LSQL) queries with streaming result support. Uses the oracle_analytics-execute_logical_sql tool. See Use the oracle_analytics-execute_logical_sql MCP Server Tool (Preview).
Data Source Types
The Oracle Analytics Cloud MCP Server allows you to access these types of Oracle Analytics Cloud data sources.
| Type | Description | Property |
|---|---|---|
| Subject Areas | Traditional subject areas generated when a semantic model is deployed to Oracle Analytics Cloud. | "Subject Area Name" |
| Extended Subject Areas (XSA) | All dataset types, for example, files, data flows, connections, and so on. | XSA('namespace'.'dataset_name') |
Prerequisites
Before you begin your development work, you must install the MCP connection utility and connect your AI client to your Oracle Analytics Cloud instance.
See Download and Install the MCP Connection Utility (Preview) and Add the Oracle Analytics Cloud MCP Server to Your AI Client (Preview).