IBM/powervs-mcp-server

IBM Power Virtual Server MCP (Model Context Protocol) Server

This repository is an example MCP (Model Context Protocol) server for IBM Power Virtual Server.

DISCLAIMER: This is a sample MCP server for reference and is not officially affiliated with, endorsed by, or supported by IBM. This MCP server utilizes the IBM Power Virtual Server External APIs.

This open-sourced Model Context Protocol (MCP) server will help IBM Power Virtual Server to integrate in the Agentic-AI ecosystem. It will help users to bring their AI-Agents for seamless observability and diagnosis of their virtual machines registered with IBM Power Virtual Server .

🚀 Features

• Observability Tools: Leverage key IBM Power Virtual Server monitoring capabilities via an MCP interface.
• Extensible Design: Easily integrate additional Power Virtual Server APIs for future expansions.
• Pythonic: Allowing ease of use and extension for AI developers

🛠️ Tools

Listed below are the tools which are presently exposed thought the MCP server:

1. fetch_powervs_vms

• Description: List all the Virtual machines from PowerVS.
• Returns: List all the virtual machines across all the workspaces and across data centers registered with the IBM account.

2. fetch_powervs_vms_by_health_status

• Description: Get Power VS virtual machines filtered by their health status.
• Inputs:
  • health_status: The probable health status are OK, CRITICAL, ERROR, WARNING, UNKNOWN.
• Returns: List all the virtual machines across all the workspaces and across data centers notifications filtered by health status.

3. fetch_powervs_critical_vms

• Description: Get all Power VS virtual machines in CRITICAL health state.
• Returns: List all the virtual machines across all the workspaces and across data centers notifications in CRITICAL health state.

4. fetch_powervs_all_workspaces

• Description: Get all Power VS workspaces for the tenant.
• Returns: List all the workspaces across all the data centers registered with the IBM account.

5. fetch_powervs_vm_network_health

• Description: Get network interface health informtion for a Power VS VM instance.
• Inputs:
  • vm_id (string): Unique system id for the virtual server.
• Returns: Details of network health of a virtual server.

6. fetch_powervs_vm_storage_health

• Description: Get storage health data for a Power VS VM instance.
• Inputs:
  • vm_id (string): Unique system id for the virtual server.
• Returns: Details of storage health of a virtual server.

7. fetch_powervs_vm_health

• Description: Get complete health details of Power VS VM including storage and network.
• Inputs:
  • vm_id (string): Unique system id for the virtual server.
• Returns: Details of holistic health of a virtual server that includes storage and network.

8. fetch_powervs_all_images

• Description: Get all available Power VS images in the workspace.
• Returns: List all Power VS images within a workspace, do pass CRN to fetch this information.

9. fetch_powervs_image_details

• Description: Get detailed information for a Power VS specific image.
• Inputs:
  • image_id (string): Unique system id for the Power VS image.
• Returns: Details of the Power VS image.

10. fetch_powervs_all_networks

• Description: Get all networks in the Power VS workspace.
• Returns: List all networks of a Power VS workspace, do pass CRN to fetch this information.

11. fetch_powervs_vm_snapshots

• Description: Get all snapshots for a Power VS VM.
• Inputs:
  • vm_id (string): Unique system id for the virtual server.
• Returns: List all the snapshots of a specifed VM.

🧪 Setup

Set up your environment

Step 1: Build the container image:

podman build -t powervs_mcp_server:latest -f Containerfile .

Step 2: Run the container:

You need to pass ACCOUNT_ID and API_KEY as mandatory attribute and CRN (workspace CRN) would be optional used only if needed to target a specific PowerVS workspace. This can either be set inside the config.yaml or pass it as runtime environment variables to podman or docker container

podman run -d --name powervs_mcp_server \
-p 8002:8002 \
-e ACCOUNT_ID=<POWERVS_ACCOUNT_ID> \
-e API_KEY=<POWERVS_API_KEY> \
-e CRN=<CRN>
localhost/powervs_mcp_server

You could also override default 8002 port by passing USER_DEFINED_PORT:

podman run -d --name powervs_mcp_server \
-p <PORT_ON_HOST>:<USER_DEFINED_PORT> \
-e MCP_SERVER_PORT=<USER_DEFINED_PORT>
-e ACCOUNT_ID=<POWERVS_ACCOUNT_ID> \
-e API_KEY=<POWERVS_API_KEY> \
-e CRN=<CRN>
localhost/powervs_mcp_server

IBM Power VS Credentials

Tools in this MCP Server invokes IBM Power VS APIs APIs and hence needs API key for a working setup. Refer Generating a IBM Cloud API Key to generate REST API key for your tenant / account ID.

Add below values to src/config.yaml file:

account_id: <IBM Cloud tentant or account ID>
api_key: <IBM Cloud API Key>
base_url: <Serivce broker URL>
crn: <IBM Power VS CRN of workspace that needs to be targeted>

🖥️ Usage with Langchain MCP Client

Here is the MCP client configuration details:

{
    "powervsserver": {
        "transport": "streamable_http",
        "url": "http://127.0.0.1:8002/mcp",
    }
}

🐞 Testing and Debugging

We recommend using the MCP Inspector for testing and debugging. You can run the inspector with:

npx @modelcontextprotocol/inspector

The inspector will provide a URL you can open in your browser to see logs and send requests manually.

📄 License

This project is licensed under the .