Create an AI Agent to manage your OAC instance With MCP and REST API's

SteveF-Oracle

Hi Community,

I am re-posting a blog, by one of our own Oracle Analytics Product Managers @Mike Durran-Oracle, that I found really fascinating.

You are able to ask questions, such as:

• What can you tell me about my Oracle Analytics instance?
• Do you think I need to create a backup?
• Are there any workbooks not owned by me?
• Give me a summary of topics my workbooks created in the last 3 months cover.
• Use lastModified to sort the most recent workbooks
• What data flow was the most recent one to execute?
• Export to a PDF, a workbook I use regularly do demo embedded analytics. This workbook has 'Dark Theme' in the name. Using actions/getCanvases, export the first canvas to PDF. Check on the status of the workrequest and give me a link that I can use to download the PDF in a tool such as Postman.
  
  

I will just quote the introduction here, then let the community dive in to explore the 'art of the possible'.

Introduction

The Model Context Protocol (MCP) was developed by Anthropic and introduced in Nov 2024. The aim is to provide a standardised way for AI systems (LLMs) to connect with other systems, such as data sources and, in turn, extending the capabilities of those AI systems. MCP is a key component of Agentic AI, where a defining aspect is the ability for the agent to act autonomously.

I recently experimented with creating my own MCP server, using a generally available version of Oracle Analytics so I could better understand this area.

In this blog, I’ll explain my process and my results with the aim of hopefully starting a discussion on this topic in relation to analytics.

https://insight2action.medium.com/experimenting-with-oracle-analytics-rest-apis-and-the-model-context-protocol-mcp-ab0cf1e165df

I hope you find it as interesting as I did. Kudos to Mike for this exploration.