fleet_bundle_list
List Fleet Bundles (deployment units from GitRepos)
LABSCO SUMMARY
Out of the box, an agent can answer questions about the estate without being able to change it.
The default launch writes nothing: --read-only is on unless you pass --read-only=false, and deletes also need --disable-destructive=false, so until you add those flags what you hand an agent is limited to what it can read. The audit log and legacy catalog calls degrade rather than fail, returning JSON that carries _source unavailable and a note when a Rancher build does not expose them. The cluster argument on the Kubernetes tools is a Rancher cluster ID.
What it is
A server over Rancher that starts read-only, reaching its Norman v3 API, the Kubernetes resources of the clusters it manages, Fleet GitOps state and Harvester virtualisation; the write and delete tools come on only with launch flags.
What you get
- Cluster inventory with health, Kubernetes version, provider and node count, plus a cross-cluster total of clusters and projects.
- Kubernetes access on a named cluster: get or list any apiVersion and kind, describe-style output with recent events, namespace event lists, pod log tails and a capacity summary drawn from Nodes.
- Fleet GitOps state — GitRepos, bundles, registered clusters, and a drift report of BundleDeployments sitting in Modified state.
- Harvester inventory: virtual machines with status, IP, node and CPU or memory, one VM in full spec, Longhorn-backed volumes, host maintenance and disk status, VM images with download progress, VLAN networks, KubeOVN subnets and VPCs, addons and cluster settings.
- Rancher administration reads: users, global role bindings, auth configs, tokens, cloud credentials, settings, feature flags, node drivers, catalogs, cluster repos and cluster registration tokens.
- Norman schema discovery, which reports the types, collection links and actions this particular Rancher exposes.
Requirements
A reachable Rancher server and an API token for it. The KubeOVN subnet and VPC tools need the kubeovn-operator addon enabled, and token and cloud credential values come back redacted unless show-sensitive-data is turned on.
Setup effort
One command plus a key — npm install -g rancher-mcp-server, then supply credentials
USE CASES
01
Working out why a pod keeps failing on a downstream cluster
List the resource, read its recent events in describe form, widen to the namespace's events, then tail the container log.
02
Catching configuration that has drifted away from Git
The drift report names the BundleDeployments changed outside GitOps; the GitRepo, bundle and cluster lists say which source and which cluster they came from.
03
Judging whether a cluster can take more workload
Allocatable resources from the Nodes, set against the cluster's own health and node count and the estate-wide totals.
04
Taking stock of the virtual machines and storage on Harvester
List the VMs with their status and placement, open the one in question in full, then check the volumes under it and the hosts they sit on.
05
Auditing who can reach Rancher and how
Users, the global role bindings that grant them scope, the auth providers they arrive through, and the API tokens standing open.
06
Tracing a VM network back to the subnet behind it
VLAN networks link to KubeOVN subnets through their provider, and the subnets in turn belong to a VPC. The subnet and VPC calls need the kubeovn-operator addon.
49 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
fleet_cluster_list
List Fleet clusters (downstream clusters registered with Fleet)
fleet_drift_detect
Report Fleet drift: BundleDeployments with Modified state (resources changed outside GitOps)
fleet_gitrepo_get
Get detailed info for one Fleet GitRepo
fleet_gitrepo_list
List Fleet GitRepos (GitOps sources)
harvester_addon_list
List Harvester addons with their enabled/disabled state
harvester_host_list
List Harvester hosts (nodes) with maintenance mode and disk status
harvester_image_list
List Harvester VM images (VirtualMachineImage) with download progress
harvester_network_list
List Harvester VLAN networks (NetworkAttachmentDefinition)
harvester_settings
List or get Harvester cluster settings (backup-target, auto-disk-provision-paths, etc.)
harvester_subnet_list
List KubeOVN Subnets (requires kubeovn-operator addon). Subnets define VM network CIDRs and are linked to Networks via provider.
harvester_vm_get
Get a single virtual machine with full spec and status (metrics, disks, networks)
harvester_vm_list
List virtual machines across Harvester clusters with status, IP, node, CPU/memory
harvester_volume_list
List PersistentVolumeClaims (Longhorn-backed volumes) with size and health
harvester_vpc_list
List KubeOVN VPCs (requires kubeovn-operator addon enabled)
kubernetes_capacity
Summarize cluster capacity and allocatable resources from Nodes
kubernetes_describe
Get a Kubernetes resource and its recent events (describe-style output)
kubernetes_events
List Kubernetes events in a namespace (optionally filtered by involvedObject)
kubernetes_get
Get a single Kubernetes resource by apiVersion, kind, namespace, name
kubernetes_list
List Kubernetes resources by apiVersion and kind (e.g. v1 Pod, apps/v1 Deployment)
kubernetes_logs
Get recent logs from a pod's container (tail only; no follow). Use for debugging. Cluster is Rancher cluster ID.
rancher_audit_log_list
Attempts GET /v3/auditlogs (Norman). Many Rancher builds return 404 (not exposed) or 405 (GET not supported); in those cases the tool returns JSON with _source unavailable, _http_status, and an explanatory _note (e.g. GET not supported for 405) instead of failing.
rancher_auth_config_get
Get an auth config by id (/v3/authconfigs/{id})
rancher_auth_config_list
List auth configs (/v3/authconfigs)
rancher_catalog_get
Get a catalog by id (/v3/catalogs/{id})
rancher_catalog_list
List legacy catalogs (Norman /v3/catalogs). On any non-success (including transport errors), returns JSON with _source unavailable instead of a hard tool error.
rancher_cloud_credential_get
Get a cloud credential by id
rancher_cloud_credential_list
List cloud credentials (/v3/cloudcredentials). Secrets redacted unless show-sensitive-data.
rancher_cluster_get
Get detailed info for one Rancher cluster (health, version, node count)
rancher_cluster_list
List all Rancher clusters with health, K8s version, provider, node count
rancher_cluster_registration_token_get
Get a cluster registration token by id
rancher_cluster_registration_token_list
List cluster registration tokens (/v3/clusterregistrationtokens). Optional cluster_id filters by cluster.
rancher_cluster_repo_get
Get a cluster repo by id. Tries Norman /v3/clusterrepos/{id} first; on 404, reads catalog.cattle.io/v1 ClusterRepo from the local cluster. Use id "namespace/name" or "name" (defaults to namespace cattle-global-data).
rancher_cluster_repo_list
List app catalog cluster repos. Tries Norman /v3/clusterrepos; on failure or non-success, lists catalog.cattle.io/v1 ClusterRepo on local via Steve/Kubernetes, trying namespaces cattle-global-data, fleet-default, fleet-local, cattle-fleet-system. Returns either _source kubernetes_api_fallback with data or _source unavailable with attempt_errors (no hard tool error). Pagination marker applies only to the first successful namespace (usually cattle-global-data).
rancher_feature_flag_get
Get a feature flag by id (/v3/features/{id})
rancher_feature_flag_list
List feature flags (/v3/features)
rancher_global_role_binding_get
Get a global role binding by id (/v3/globalrolebindings/{id})
rancher_global_role_binding_list
List global role bindings (/v3/globalrolebindings)
rancher_node_driver_list
List node drivers (/v3/nodedrivers)
rancher_norman_schema_get
Get a single Norman schema by id (e.g. user, token, cluster) from /v3/schemas/{id}
rancher_norman_schema_list
List Norman /v3 API schemas (discover types, collection links, and actions for this Rancher)
rancher_overview
Cross-cluster summary: total cluster count and total project count
rancher_project_list
List Rancher projects across clusters
rancher_setting_get
Get a setting by id (/v3/settings/{id})
rancher_setting_list
List global Rancher settings (/v3/settings)
rancher_token_get
Get a Norman token by id (/v3/tokens/{id}). Token value redacted unless show-sensitive-data is enabled.
rancher_token_list
List Norman API tokens (/v3/tokens). Sensitive token values are redacted unless show-sensitive-data is enabled.
rancher_user_get
Get a user by id (/v3/users/{id})
rancher_user_list
List Rancher users (/v3/users)
BEFORE YOU INSTALL
What to know first
Tools you get
49 tools when we connected; the project documents 98. --read-only defaults to true and --toolsets to harvester only. Add --read-only=false for writes, --disable-destructive=false for deletes, and --toolsets harvester,rancher,kubernetes,helm,fleet for the rest.
Requires
- npx on PATH (or Docker for the HTTP route)
- A Rancher server URL and API token, passed as --rancher-server-url / --rancher-token or RANCHER_MCP_RANCHER_SERVER_URL / RANCHER_MCP_RANCHER_TOKEN
INSTALLATION
Connect your client
Server setup
TERMINAL
npm install -g rancher-mcp-serverMCP Clients
Cursor, Claude Desktop
JSON
{
"mcpServers": {
"rancher": {
"command": "npx",
"args": [
"-y",
"rancher-mcp-server",
"--rancher-server-url",
"<YOUR_RANCHER_MCP_RANCHER_SERVER_URL>",
"--rancher-token",
"<YOUR_TOKEN>:yyyy",
"--toolsets",
"harvester,rancher,kubernetes,fleet"
]
}
}
}Claude Code
Run the server and expose the port
CLAUDE CODE · TERMINAL
docker run -d -p 8080:8080 \
-e RANCHER_MCP_RANCHER_SERVER_URL=<YOUR_RANCHER_MCP_RANCHER_SERVER_URL> \
-e RANCHER_MCP_RANCHER_TOKEN="<YOUR_TOKEN>:yyyy" \
-e RANCHER_MCP_TRANSPORT=http \
-e RANCHER_MCP_PORT=8080 \
ghcr.io/mrostamii/rancher-mcp-server:latestConnect Claude Code to the MCP endpoint
The server answers at http://localhost:8080/mcp (Streamable HTTP; the default path is /mcp).
CLAUDE CODE · TERMINAL
claude mcp add -t http rancher http://localhost:8080/mcp
