Must-Gather MCP Server

A Model Context Protocol (MCP) server that provides AI assistants with comprehensive access to OpenShift must-gather data for troubleshooting and diagnostics.

Overview

The must-gather MCP server enables AI assistants to deeply analyze OpenShift clusters through must-gather archives, providing:

• Complete cluster configuration and status
• Network connectivity and performance data
• ETCD database health and metrics
• Pod and node diagnostics with logs
• Operator status and version tracking

Features

🚀 Performance

• Fast Resource Access: In-memory indexing for ~11,000 resources
• Quick Startup: 5-10 seconds to load and index
• Fast Queries: <50ms for indexed resource lookups
• On-Demand Logs: Logs loaded only when requested

🛠️ Tool Categories (62 Tools Across 5 Toolsets)

Cluster Toolset (28 tools)

Version & Info (6 tools):

• cluster_version_get - OpenShift version, update status, capabilities
• cluster_info_get - Infrastructure (platform, region, topology, network config)
• cluster_operators_list - All operators with Available/Progressing/Degraded status
• cluster_operator_get - Detailed operator conditions, versions, related objects
• cluster_nodes_list - Nodes with roles, status, kubelet version
• cluster_node_get - Detailed node info (capacity, conditions, taints)

Machine Configuration (4 tools):

• machineconfig_list - List all MachineConfigs (OS and systemd configuration)
• machineconfig_get - Get detailed MachineConfig including ignition configuration
• machineconfigpool_status - Check pool degradation and update status
• machineconfignode_status - Per-node config application status

Storage (4 tools):

• storage_classes_list - List StorageClasses with provisioners and defaults
• csi_drivers_status - CSI driver capabilities and status
• volume_attachments_list - Find stuck or failing volume mounts
• persistent_volumes_status - PV status (available, bound, failed)

Security & RBAC (2 tools):

• security_scc_list - SecurityContextConstraints with privilege analysis
• rbac_clusterroles_list - ClusterRoles with dangerous permission detection

OLM/Operators (3 tools):

• olm_subscriptions_status - Operator subscription and upgrade status
• olm_catalogsources_status - Catalog source connectivity and health
• olm_installplans_status - Pending operator installations

Admission Control (2 tools):

• admission_webhooks_list - Validating and mutating webhooks
• admission_policies_list - CEL-based admission policies

Configuration (2 tools):

• cluster_config_list - List all config.openshift.io resources
• cluster_config_get - Get detailed cluster configuration

Ingress & Routing (4 tools):

• routes_list - List OpenShift Routes with hosts, services, TLS config, and admission status
• route_get - Get detailed Route information including backend weights and TLS termination
• ingress_list - List Kubernetes Ingress resources with rules and backends
• ingresscontroller_status - Get IngressController (router) status, replicas, and conditions

Core Toolset (6 tools)

Resources (3 tools):

• resources_get - Get any Kubernetes resource by kind/name/namespace
• resources_list - List resources with label/field selectors
• namespaces_list - List all namespaces

Events (3 tools):

• events_list - Filter events by type/namespace/reason
• events_timeline - Chronological event sequence for incident analysis
• events_by_resource - All events for specific pods/nodes/deployments

Diagnostics Toolset (17 tools)

Pod Logs (2 tools):

• pod_logs_get - Container logs (current/previous) with tail support
• pod_containers_list - Discover containers with logs

Node Diagnostics (4 tools):

• nodes_list - Nodes with diagnostic data available
• node_diagnostics_get - Comprehensive node diagnostics (kubelet, sysinfo, CPU/IRQ, hardware)
• node_kubelet_logs - Kubelet logs (auto-

... View full README on GitHub