Managing modern cloud environments often means juggling numerous commands, remembering syntax variations, frequent context-switching between consoles, and dealing with a high risk of operator errors—all of which slow down productivity and increase the risk of mistakes. IBM addresses these challenges by integrating the power of the IBM Cloud CLI with the intelligence of the IBM Cloud MCP (Model Context Protocol) Server, evolving the CLI into a smart, conversational, agentic AI assistant for the cloud.

What is IBM Cloud MCP?

IBM Cloud MCP extends the IBM Cloud CLI by enabling users to issue natural-language requests instead of precise commands. Operating as a local install or a containerized service, MCP interprets conversational instructions and translates them into secure CLI operations across resources like VPCs, Kubernetes clusters, IAM policies, and Object Storage. It operates with either a local, single-user install or in containerized deployments, supporting MCP-compliant clients and large language models (LLMs).

Administrators have full control over which CLI tools MCP can access, maintaining strict security boundaries. By default, MCP runs in read-only mode, enforcing the principle of least privilege—write access must be explicitly enabled. This approach ensures security and relevance for each use case.

All MCP features are provided as EXPERIMENTAL features at this time.

What is IBM Cloud CLI?

The IBM Cloud CLI is a command-line interface that enables users to interact programmatically and directly with IBM Cloud services. It supports a wide range of cloud operations:

• Core infrastructure management (VPCs, Kubernetes clusters, networking)
• IAM and access control (policies, roles)
• Automation workflows and advanced scripting
• Database and Object Storage services
• Plugin extensibility and support for both ad hoc operations and automated DevOps workflows

Deployment Options and Setup:

1. Local Single-user Install: Install the IBM Cloud CLI, configure an MCP-compliant host application (e.g., VSCode, Cline, mcp-cli), and set MCP options (such as --mcp-transport stdio, --mcp-tools). Useful for individuals who want conversational management on their local systems.

2. Containerized Deployment: Use the provided Containerfile or makefile to build and deploy MCP servers on any container runtime (Docker, Podman, Kubernetes, or IBM Cloud Code Engine). Secure each deployment with a dedicated service identity and expose only the necessary tools. Deploy using the included Makefile scripts, following specific instructions from the open-source repository.

Use Cases:

The integration of conversational AI transforms the IBM Cloud CLI by bridging manual syntax with plain-language, agentic automation. Below are several prominent examples in major core services of IBM Cloud:

     1.   Cloud Kubernetes and Openshift

Sample prompts:

1. Show quotas grouped by Infrastructure and grouped by Type, then sort by region.and quota amount in ascending order.
2. Show a dashboard with the operational status of the storage volumes by geography. Only show vpc-gen2 storage devices.

     2.   PowerVS

Sample prompts:

1. List the instances in this workspace [in table format]
2. Create an instance named {{INSTANCE_NAME}} in the {{WORKSPACE_NAME}} workspace, with the latest {{IMAGE_NAME}} operating system, and minimal cpu/memory profile attached to {{SUBNET_NAME}}

     3.   Container Registry

Sample prompts:

1. Show me the details of issue {{ISSUE-NAME}}
2. Show vulnerability report for IMAGE_NAME. Include suggestions for remediation actions.

Benefits of Conversational AI in Cloud Management:

• Reduces Error and Complexity: Interprets user intent, runs validated commands, and reduces syntax mistakes.
• Boosts Productivity: Removes the burden of memorizing command syntax and consulting documentation, speeding up routine workflows.
• Empowers Non-experts: Makes cloud management accessible to those with less CLI background, lowering entry barriers.
• Automates Routine and Complex Tasks: Takes on everything from simple resource queries to complex policy configuration in a single conversational request

Limitations and Security:

• MCP is currently designed for single-user or single-service scenarios and stores configurations locally.
• OAuth authentication is not supported.
• Security best practices advise minimal access privileges, segregated APIs, and trusted agents to prevent unauthorized access or data leaks.

Getting Started with IBM Cloud MCP:

1. Local Setup (on your computer)

1. Install the IBM Cloud CLI (follow this guide).
2. Check if MCP features are available.
3. If you don’t see MCP, update your IBM Cloud CLI.
4. Get your IBM Cloud API Key (or log in with ibmcloud login –sso).
5. Log in using the API key.
6. Install corresponding plugins and configurations to ensure that MCP host knows about the serv
7. Find the path to your IBM Cloud CLI.
8. Configure your host app to use MCP.
9. (Optional) To allow creating/deleting resources, add --mcp-allow-write to the "args" list in the host app’s JSON.

2. Container Setup (Docker / Podman / Code Engine)

1. Install Docker or Podman and make.
2. Clone the MCP repository.
3. Build the container.
4. Create a .env file with your IBM Cloud API key, region, and group.
5. Run the container.

3. Deploy on IBM Cloud Code Engine

1. Create a .env.ce file with your API key, region, and Code Engine settings.
2. Verify your environment variables.
3. Install IBM Cloud CLI + plugins (if not already).
4. Log in to IBM Cloud.
5. Select your Code Engine project.
6. Build and tag the image (if not already).
7. Push the image to IBM Container Registry.
8. Deploy or update the app in Code Engine.
9. Monitor and manage the app.

Conclusion:

IBM Cloud MCP transforms the command-line interface into a context-aware, intelligent assistant for cloud management, delivering improvements in operational efficiency, error reduction, and automation accessibility. The CLI becomes not just a tool, but an intelligent conversational partner, paving the way for even more autonomous workflows and integrations as the platform evolves. The journey from commands to conversations has begun, and with IBM Cloud MCP, the future is agentic, efficient, and accessible.

References:

1. https://ibm-cloud.github.io/mcp/overview/
2. https://cloud.ibm.com/docs/cli
3. https://cloud.ibm.com/docs/containers
4. https://cloud.ibm.com/docs/account?topic=account-iamoverview
5. https://cloud.ibm.com/docs/cloud-object-storage
6. https://cloud.ibm.com/docs/account?topic=account-serviceidapikeys&interface=ui
7. https://cloud.ibm.com/docs/cli?topic=cli-install-ibmcloud-cli
8. https://www.ibm.com/products/code-engine
9. https://ibm-cloud.github.io/mcp/servers/container-registry/
10. https://ibm-cloud.github.io/mcp/servers/kubernetes/
11. https://ibm-cloud.github.io/mcp/servers/powervs/