Conectar

Open Cluster Management

ocm-mcp-serverbyocm-mcp-server
•
16
•
Infraestructura en la Nube
Despliegue y DevOps
Seguridad y Pruebas

Operate multi-cluster Kubernetes fleets through Open Cluster Management with policy guardrails, human approvals, and comprehensive auditing.

ocm-mcp-server acts as a secure AgentOps control plane between AI models and Kubernetes fleets via Open Cluster Management (OCM). Instead of granting models direct cluster-admin access or raw kubeconfigs, the server implements a four-layer defense system: static validation rules, Kyverno policy admission dry-runs, cryptographic Ed25519 human approval tokens, and least-privilege Kubernetes RBAC. This allows AI assistants across multiple platforms to safely diagnose fleet health, propose workload deployments as ManifestWorks, and execute remediations without risking unverified, destructive changes.

Características Principales

01Cryptographic human-in-the-loop approvals using Ed25519 one-time tokens bound to content and operation
02Zero-kubeconfig access for AI agents to prevent direct or unrestricted cluster manipulation
0316 GitHub stars
04Multi-cluster operations through Open Cluster Management primitives like ManagedCluster and ManifestWork
05Complete audit logging and OpenTelemetry tracing for all read actions, proposals, and applied changes
06Four-layer security model enforcing static checks, Kyverno policies, human approval, and RBAC

Casos de Uso

01Executing safe automated remediations that require explicit human approval signatures before deployment
02Safely diagnosing degraded workloads across multi-cluster environments using free read queries
03Proposing and validating fleet-wide configuration changes against organizational Kyverno policies