Kubeleash
Give coding agents Kubernetes operations with RBAC-style, context-scoped guardrails so prod changes stay bounded.
Overview
io.github.kubeleash/kubeleash is an MCP server for the Operate phase that lets AI agents interact with Kubernetes under RBAC-style, context-scoped guardrails.
What is this MCP server?
- Kubernetes MCP server designed for AI agent callers
- RBAC-style permissions with context-scoped guardrails
- Shipped as OCI image ghcr.io/kubeleash/kubeleash:0.3.0 with stdio transport
- Reduces risk of unconstrained kubectl-style agent actions in shared clusters
- Version 0.3.0 on registry
- OCI identifier ghcr.io/kubeleash/kubeleash:0.3.0
- RBAC-style context-scoped guardrails per description
Community signal: 1 GitHub stars.
What problem does it solve?
Letting agents touch Kubernetes without tight scopes is terrifying—you need helper tooling, not a free-range cluster admin.
Who is it for?
Indie operators running small K8s fleets who want agent-assisted triage, inspection, or bounded changes with explicit policy.
Skip if: Builders on serverless-only or docker-compose stacks, or teams unwilling to configure RBAC and audit agent actions.
What do I get? / Deliverables
Agents can perform allowed cluster operations through MCP while RBAC-style guardrails limit blast radius to the contexts you define.
- Agent-callable Kubernetes tools limited by context-scoped guardrails
- Reduced need for hand-written kubectl scripts in incident chat
- Documented OCI-based stdio MCP deployment path
Recommended MCP Servers
Journey fit
Production Kubernetes control belongs in operate once you are running clusters, not during first prototype wiring. Infra subphase fits cluster-scoped agent tools with policy guardrails rather than pre-ship unit tests or launch SEO.
How it compares
Guardrailed Kubernetes MCP bridge, not a passive traffic analyzer like Kubeshark or a generic cloud cost dashboard.
Common Questions / FAQ
Who is io.github.kubeleash/kubeleash for?
Solo builders and small teams operating Kubernetes who want MCP-connected agents with scoped permissions instead of full admin kubeconfig in chat.
When should I use io.github.kubeleash/kubeleash?
Use it during operate and infra work when agents need read or limited write cluster access under RBAC-style guardrails you have reviewed.
How do I add io.github.kubeleash/kubeleash to my agent?
Pull the OCI package ghcr.io/kubeleash/kubeleash:0.3.0, configure stdio MCP in your client, mount kubeconfig or in-cluster credentials, and apply kubeleash RBAC or context scopes per the GitHub repository.