MCP-OpenStack-Ops

MCP-OpenStack-Ops is an MCP server that connects LLM clients to OpenStack cloud environments for infrastructure management and cluster monitoring. Built with Python and the OpenStackSDK, it allows DevOps engineers, cloud administrators, and infrastructure operators to interact with OpenStack resources through conversational AI interfaces like Claude Desktop. The server strictly enforces project-scoped operations tied to the OS_PROJECT_NAME environment variable, preventing cross-tenant access. It provides over 90 purpose-built tools spanning compute (Nova), networking (Neutron), block storage (Cinder), orchestration (Heat), identity (Keystone), and load balancing (Octavia). Safety controls keep destructive or mutating operations gated behind an explicit ALLOW_MODIFY_OPERATIONS environment toggle, ensuring read-only safety by default. Users can query compute instances, check cluster utilization quotas, monitor amphora load balancers, and execute targeted bulk lifecycle tasks using name-based pattern matching.

Category: Cloud & Infrastructure

Tags: cloud, cluster-management, infrastructure, monitoring, openstack

Visit MCP-OpenStack-Ops

How to install and configure MCP-OpenStack-Ops

To install and configure MCP-OpenStack-Ops, refer to the project repository for complete setup instructions: 1. Clone the repository from https://github.com/call518/MCP-OpenStack-Ops. 2. Configure your OpenStack environment variables, including OS_PROJECT_NAME and ALLOW_MODIFY_OPERATIONS. 3. If using an older release like Wallaby, ensure the openstacksdk version matches in Dockerfile.MCP-Server and pyproject.toml. 4. Build and run the container using Docker Compose with docker-compose build --no-cache mcp-server and docker-compose up -d. 5. Register the running container or Python process within your MCP client configuration (such as Claude Desktop) using standard stdio or HTTP transports as detailed in the README.

What you can do with MCP-OpenStack-Ops

• Managing instance lifecycles by issuing bulk start, stop, reboot, resize, or delete actions targeted through substring filters. - Auditing OpenStack infrastructure health, tenant quota limits, service availability, and resource utilization statistics across project workloads. - Administering Octavia load balancers, listeners, pools, health monitors, and member backends under single-tenant safety boundaries. - Orchestrating block storage operations including volume attachment, detachment, snapshot creation, and quota tracking. - Attaching and detaching floating IPs, security groups, and virtual network interfaces for compute instances safely.

Key facts

• Open Source
• https://github.com/call518/MCP-OpenStack-Ops
• Cloud & Infrastructure, Monitoring & Observability
• cloud, cluster-management, infrastructure, monitoring, openstack

Part of MCP Servers

What is MCP-OpenStack-Ops?

MCP-OpenStack-Ops is a Model Context Protocol server that connects language models to OpenStack clouds. It enables users to view, manage, and monitor infrastructure resources including virtual machines, block volumes, networks, and load balancers through natural language prompts within a specified project tenant.

Which OpenStack versions are supported by MCP-OpenStack-Ops?

MCP-OpenStack-Ops is primarily tested and optimized for OpenStack Epoxy (2025.1) and Dalmatian (2024.2). Older releases such as Caracal, Bobcat, and Wallaby can also be used by matching the openstacksdk dependency version across Dockerfile.MCP-Server and pyproject.toml as documented in the repository.

How does MCP-OpenStack-Ops handle write safety?

All mutating or modifying operations, such as creating or deleting servers and volumes, are disabled by default. These tools are only registered when the ALLOW_MODIFY_OPERATIONS environment variable is explicitly set to true, preventing unintended infrastructure modifications during read-only diagnostic sessions.

Is MCP-OpenStack-Ops open source?

Yes, MCP-OpenStack-Ops is open source software distributed under the MIT license. You can freely inspect the code, modify it for your specific private cloud architecture, and run it locally via Docker or Python.