Kubernetes MCP
Free foreverstrowk · Dev Tools
18k installs
Kubernetes MCP gives AI agents direct, structured visibility into your cluster state so they can answer operational questions, triage incidents, and assist with day-to-day Kubernetes cluster management without a human copy-pasting kubectl output into a chat window. Instead of manually inspecting pods, services, and deployments, an agent connected through this MCP server can query live cluster data on demand — namespace by namespace, resource by resource — and reason about what's actually running versus what's expected. This makes it a natural fit for teams who want AI-assisted DevOps: faster root-cause analysis when a deployment is stuck, quicker answers about which service is exposing which port, and always-current context for anyone (human or agent) trying to understand the shape of a cluster.
On BusinessMCP, this Kubernetes MCP server is hosted alongside the rest of your company's tools, databases, ad platforms, and revenue data inside a single unified MCP endpoint. Rather than standing up and maintaining a bespoke Kubernetes MCP integration per agent or per client, you connect your cluster credentials once through our managed hosting layer, and every downstream AI agent — Claude, GPT, Gemini, or any model-agnostic client — reaches cluster state through the same /api/mcp endpoint using a Bearer mcph_* key. That means your Kubernetes tooling sits next to your Docker, CI/CD, observability, and infrastructure data in one place, so an agent can correlate a failing deployment with a recent CircleCI build, a Datadog alert, or a Terraform apply without you wiring together five separate MCP servers by hand.
Because BusinessMCP layers a business-intelligence dashboard on top of every connected MCP server, Kubernetes cluster activity isn't just queryable by agents — it's also visible to the humans running the growth suite. Platform teams get a consolidated view of what tools are being queried, how often, and by which workflows, which is useful when Kubernetes is one piece of a larger DevOps and revenue-facing stack. The hosting is cookieless and GDPR-friendly by design, so cluster inspection tools can be exposed to AI agents in customer-facing or internal support contexts without adding compliance overhead.
Typical use is DevOps-first: SRE and platform teams giving an AI assistant read access to pods, services, and deployments for faster incident response; engineering leads asking natural-language questions about cluster state during a rollout; support or on-call agents that need to check
Just say it in a thread
No configs, no docs. Once connected, these are the kinds of messages your agents act on.
"Return pods across one or more namespaces, including status and restart counts — and give me the highlights."
"Fetch the current rollout status and replica counts for a named deployment for me, then post a summary in the thread."
"List services in a namespace along with their type, ports, and selectors and flag anything that needs my approval."
What teams use it for
- An SRE asks an AI agent to list all failing pods across namespaces during an incident
- A support agent checks whether a customer-facing deployment rolled out successfully before replying to a ticket
- An engineering lead queries current service and deployment counts as part of a release readiness review
- An AI coding assistant cross-references Kubernetes state with recent CI builds to explain a broken deployment
- A platform team exposes read-only cluster visibility to multiple internal AI tools through one hosted endpoint
Agent-callable tools
list_pods
Return pods across one or more namespaces, including status and restart counts.
get_deployment_status
Fetch the current rollout status and replica counts for a named deployment.
list_services
List services in a namespace along with their type, ports, and selectors.
describe_cluster_state
Summarize overall cluster health, node status, and resource utilization.
get_pod_logs
Retrieve recent log output for a specified pod and container.
list_namespaces
Return all namespaces currently present in the connected cluster.
get_events
Fetch recent Kubernetes events relevant to a namespace or resource for troubleshooting.
Your data stays yours
Credentials live in your vault. We route requests — we never store, log, or train on your data.
Works with every AI
Connect once — portable across Claude, GPT, Gemini, and every local agent you run.
Pairs well with
Best Dev Tools MCP servers
Kubernetes Manager
flux159
Full Kubernetes management including deployments, services, pods, logs, and cluster operations.
Docker
docker
Manage Docker containers, images, volumes, and networks through MCP. Build, run, and monitor containerized applications.
Docker Manager
ckreiling
Full Docker container lifecycle management. Build images, run containers, manage networks and volumes.
Terraform
hashicorp
Infrastructure as code management. Plan, apply, and manage cloud infrastructure declaratively.
CircleCI
nicekid1
CI/CD pipeline management. Trigger builds, view workflow status, and manage project settings.
Frequently asked questions
Does Kubernetes MCP let agents modify cluster resources, or is it read-only?
The core exposed tools focus on retrieving pods, services, deployments, and cluster state; write operations depend on the permissions granted to the connected credentials, so most teams start with read-only access for safety.
How does this connect to my existing cluster?
You provide cluster access credentials once through BusinessMCP's managed setup, and the server exposes that cluster's state through your unified /api/mcp endpoint using a Bearer mcph_* key.
Can this Kubernetes MCP server work alongside my other DevOps tools?
Yes — because it's hosted in the same unified MCP endpoint as your other connected servers, an agent can query Kubernetes state alongside Docker, CI/CD, and observability data in a single session.
Keep exploring
Give your AI team the Kubernetes MCP skill
Free forever plan, no credit card. Connected and working in under five minutes.
Connect Kubernetes MCP free