oci-mcp-server

What is oci-mcp-server?

oci-mcp-server is a Python MCP (Model Context Protocol) server for Oracle Cloud Infrastructure (OCI) API Gateway management. It provides tools to list and retrieve details of API Gateways, and supports both programmatic clients and LLM-powered agents (via stdio transport, Cursor, and LlamaIndex/Ollama).

How to use oci-mcp-server?

Install Python 3.11, the uv package manager, and create a virtual environment. Install dependencies from pyproject.toml. Configure OCI compartment, gateway OCIDs, region, and profile in src/config.py, and ensure valid OCI session authentication files exist. Run python src/server.py to start the server via stdio transport. For Cursor integration, edit ~/.cursor/mcp.json using the template in src/mcp_template.json. Optionally run the programmatic client (python src/client_stdio.py) or the WIP LlamaIndex/Ollama client (python src/client_llama_react.py).

Key features of oci-mcp-server

• List API Gateways in a given OCI compartment.
• Get details for a specific API Gateway.
• MCP stdio transport for easy integration.
• LlamaIndex/Ollama client for natural language interaction.
• Programmatic client for testing and debugging.

Use cases of oci-mcp-server

• Manage OCI API Gateways using an AI-powered code editor (Cursor).
• Automate API Gateway inventory and monitoring tasks.
• Build custom automation scripts with programmatic access.
• Experiment with self-hosted LLM agents for OCI resource management.

FAQ from oci-mcp-server

What is oci-mcp-server?

oci-mcp-server is a Python MCP server that exposes tools to list and retrieve details of OCI API Gateways, usable from both traditional code and AI agents.

What are the prerequisites?

Python 3.11, the uv package manager, an OCI account with API Gateway access, and OCI session authentication files located in ~/.oci/sessions/DEFAULT/.

What tools does the server expose?

Two tools: list_gateways_tool(compartment_id) and get_gateway_tool(gateway_id).

How do I integrate with Cursor?

Update the mcp.json file with absolute paths to your Python executable and server.py, following the example in src/mcp_template.json.

Is the LlamaIndex/Ollama client fully functional?

No, it is a work in progress (WIP). Ensure Ollama is running locally with the llama3.2 model before using it.