MCP Oci Integration

🚀 MCP Oracle OCI Integrations

This repository provides code and examples to facilitate the following tasks:

• Develop MCP servers using Python.
• Run MCP servers on Oracle OCI.
• Integrate MCP servers with AI Agents.
• Integrate MCP servers with other OCI resources (ADB, Select AI, etc.).
• Integrate MCP Servers running on OCI with AI Assistants such as ChatGPT, Claude.ai, and MS Copilot.
• Integrate MCP Servers with OCI APM for Observability.
• Guide on creating a docker image for your MCP server.

🚀 Quick Start

This repository offers a wealth of resources to assist you in developing, running, and integrating MCP servers with various services and tools. Dive in and start leveraging the power of MCP!

✨ Features

What is MCP?

MCP (Model Context Protocol) is an open - source standard that enables AI models (e.g., LLMs or agents) to establish bidirectional connections with external tools, data sources, and services through a unified interface.

It addresses the “N×M” integration problem (where custom code is required for each AI - data source combination) by introducing a single standard protocol.

MCP supports dynamic discovery of available tools and context, allowing:

• AI Assistants to access relevant information from the Enterprise Knowledge base.
• Agents to reason and chain actions across different systems.

It's rapidly gaining popularity, with major players like OpenAI, Google DeepMind, and Oracle adopting it to enhance the composability and interoperability of AI systems.

In the current landscape of agentic AI, MCP is crucial as it enables models to operate effectively in real - world systems instead of remaining isolated black boxes.

Develop MCP Servers in Python

The simplest approach is to utilize the FastMCP library.

Examples:

• In Minimal MCP Server, you'll find an excellent, minimal example of a server that exposes two tools, with the option to secure them using JWT.

If you prefer to start with something more straightforward, refer to how to start developing MCP. It's simpler and doesn't support JWT tokens.

How to test

If you wish to quickly test the MCP server you've developed (or the minimal example provided here), you can use the Streamlit UI.

In the Streamlit application, you can:

• Specify the URL of the MCP server (the default is in mcp_servers_config.py).
• Select one of the models available in OCI Generative AI.
• Test asking questions and getting answers using the tools exposed by the MCP server.

In llm_with_mcp.py, there's a complete implementation of the tool - calling loop.

Semantic Search

This repository contains a complete implementation of an MCP server that performs Semantic Search on top of Oracle 23AI. To use it, you only need to:

• Load the documents into the Oracle DB.
• Place the correct configuration for connecting to the DB in config_private.py.

The code is available here.

Access to Oracle 23AI Vector Search is through the new langchain - oci integration library

Adding security

If you plan to deploy your MCP server in a production environment, you need to enhance security at multiple levels.

Here are some important points:

• Avoid exposing the MCP server directly over the Internet.
• Ensure that communication with the MCP server is encrypted (e.g., using TLS).
• Authenticate and authorize clients.

Using OCI services, you can take several steps to achieve the appropriate level of security:

• Place an OCI API Gateway in front and use it for TLS termination.
• Enable authentication using JWT tokens.
• Use OCI IAM to generate JWT tokens.
• Leverage OCI network security.

More details can be found on a dedicated page.

Integrate MCP Semantic Search with ChatGPT

If you deploy the MCP Semantic Search server, you can test its integration with ChatGPT in Developer Mode. It provides a search tool that complies with OpenAI specifications.

Soon, we'll introduce a server fully compliant with OpenAI specifications, which can be integrated into Deep Research. The server must implement two methods (search and fetch) with different behaviors, strictly following OpenAI specs.

An initial implementation is available here

Details are available here

Integrate OCI ADB Select AI

Another option is to use an MCP server to integrate OCI SelectAI with ChatGPT or other assistants that support MCP. In this way, you can perform full Text2SQL searches on your database schema. Then, the AI assistant can process the retrieved data.

An example is here

For Select AI configuration, refer to here