OpenStack

What is OpenStack MCP Server?

OpenStack MCP Server is an OpenStack resource query service that leverages the Model Context Protocol (MCP) to provide API interfaces for interacting with OpenStack cloud platform resources. It allows users to query compute, storage, network, and image resources from OpenStack.

How to use OpenStack MCP Server?

To use OpenStack MCP Server, you first need to install it via pip:

pip install openstack-mcp-server

Then, start the service by providing your OpenStack API endpoint and credentials:

openstack-mcp-server --port 8000 --log-level INFO --auth-url 'http://<OpenStack-API-Endpoint>:5000/v3' --username '<OpenStack-Admin-User>' --password '<OpenStack-Admin-Password>'

The MCP interface will be available at http://localhost:8000/openstack. You can then use tools that support the MCP protocol to query OpenStack resources, for example, to get OpenStack VM instances:

{
  "name": "get_instances",
  "arguments": {
    "filter": "web-server",
    "limit": 10,
    "detail_level": "detailed"
  }
}

Key Features of OpenStack MCP Server

• Real-time Resource Queries: Obtain the latest resource status of OpenStack clusters via API.
• Multi-dimensional Information: Support querying various resources including compute, storage, network, and images.
• Flexible Filtering: Filter resources by name, ID, and other conditions.
• Detail Level Control: Support basic, detailed, and full information display levels.
• Standard MCP Interface: Fully compatible with MCP protocol, enabling seamless integration with large language models.
• High-performance asynchronous HTTP service: Built on Starlette and Uvicorn.
• OpenStack SDK Integration: Interacts with OpenStack API using OpenStack SDK.
• SSE streaming output: For real-time feedback.

Use Cases of OpenStack MCP Server

• Integrating OpenStack with AI Clients/LLMs: Provides a standardized way for AI clients and large language models to interact with and query OpenStack resources.
• Automated Cloud Management: Enables programmatic access to OpenStack resource information for automation scripts and tools.
• Real-time Monitoring: Facilitates real-time monitoring of OpenStack cluster resources.
• Custom Dashboards and Applications: Can be used as a backend for building custom dashboards or applications that display OpenStack resource information.

FAQ about OpenStack MCP Server

Q: What is MCP? A: MCP stands for Model Context Protocol, which is a protocol designed to facilitate the interaction between large language models and external tools or services.

Q: What OpenStack resources can I query? A: You can query compute, storage, network, and image resources.

Q: Can I filter the query results? A: Yes, you can filter resources by name, ID, and other conditions.

Q: What are the detail levels for returned information? A: You can choose between basic, detailed, and full information display levels.

Q: How can I add new resource query tools? A: You can add corresponding resource acquisition functions in src/mcp_openstack_http/server.py, register the new tool in the list_tools method, and implement the tool's processing logic in the call_tool method.

OpenStack MCP Server

Overview

An OpenStack resource query service based on MCP (Model Context Protocol), providing API interfaces to query compute, storage, network, image, and other resources from the OpenStack cloud platform.

Features

• Real-time Resource Queries: Get the latest resource status of OpenStack clusters via API
• Multi-dimensional Information: Support querying various resources including compute, storage, network, and images
• Flexible Filtering: Filter resources by name, ID, and other conditions
• Detail Level Control: Support basic, detailed, and full information display levels
• Standard MCP Interface: Fully compatible with MCP protocol, seamlessly integrating with large language models

Technical Architecture

• High-performance asynchronous HTTP service based on Starlette and Uvicorn
• Interact with OpenStack API using OpenStack SDK
• Provide structured OpenStack resource information to large language models through MCP protocol
• Support SSE streaming output for real-time feedback

Architecture Diagram

┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│             │    │             │    │             │
│  AI Client  │───▶│  MCP Server │───▶│  OpenStack  │
│  (LLM)      │◀───│  (Server)   │◀───│  API        │
│             │    │             │    │             │
└─────────────┘    └─────────────┘    └─────────────┘

Installation

Requirements

• Python 3.10+
• OpenStack environment

Install via pip

pip install openstack-mcp-server

Usage

Start the Service

openstack-mcp-server --port 8000 --log-level INFO --auth-url 'http://<OpenStack-API-Endpoint>:5000/v3' --username '<OpenStack-Admin-User>' --password '<OpenStack-Admin-Password>'

After starting, the MCP interface will be available at http://localhost:8000/openstack.

Parameters

• --port: Service listening port, default is 8000
• --log-level: Log level, options are DEBUG, INFO, WARNING, ERROR, CRITICAL, default is INFO
• --json-response: Use JSON response instead of SSE stream, default is False

Interface Example

Through the MCP protocol, you can use the following tools to query OpenStack resources:

Get OpenStack VM Instances

{
  "name": "get_instances",
  "arguments": {
    "filter": "web-server",
    "limit": 10,
    "detail_level": "detailed"
  }
}

Parameter description:

• filter: Filter condition, such as instance name or ID (optional)
• limit: Maximum number of results to return (optional, default 100)
• detail_level: Level of detail in the returned information, options are basic, detailed, full (optional, default detailed)

Install from Source

# Clone repository
git clone https://github.com/wangshqly0407/openstack-mcp-server.git
cd openstack-mcp-server
# Create virtual environment
uv venv
# Activate virtual environment
source .venv/bin/activate
# Initialize runtime environment
uv sync
# Start streaming HTTP MCP server
uv run ./src/mcp_openstack_http/server.py --port 8000 --log-level INFO --auth-url 'http://<OpenStack-API-Endpoint>:5000/v3' --username '<OpenStack-Admin-User>' --password '<OpenStack-Admin-Password>'

Testing and Verification

# Method 1: Test using npx
npx -y @modelcontextprotocol/inspector uv run ./src/mcp_openstack_http/server.py --port 8000 --log-level INFO --auth-url 'http://<OpenStack-API-Endpoint>:5000/v3' --username '<OpenStack-Admin-User>' --password '<OpenStack-Admin-Password>'

# Method 2: Test using docker
docker run -it --rm -p 6274:6274 -p 6277:6277 -v $(pwd):/app -w /app node:18 npx -y @modelcontextprotocol/inspector uv run ./src/mcp_openstack_http/server.py --port 8000 --log-level INFO --auth-url 'http://<OpenStack-API-Endpoint>:5000/v3' --username '<OpenStack-Admin-User>' --password '<OpenStack-Admin-Password>'

Access: http://localhost:6274/

Development Extensions

Adding New Resource Query Tools

1. Add corresponding resource acquisition functions in src/mcp_openstack_http/server.py
2. Register the new tool in the list_tools method
3. Implement the tool's processing logic in the call_tool method

License

Apache 2.0