pve-mcp-server
Provides read-only tools for inspecting Proxmox VE resources, including VMs, LXC containers, nodes, storage, snapshots, cluster resources, backups, HA status, replication jobs, firewall rules, and more.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@pve-mcp-server List all virtual machines on the cluster"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
pve-mcp-server
Read-only Proxmox VE MCP server — 25 tools for VM/LXC/node/storage/cluster inspection. Single cluster per instance, stdio transport only.
Config
Copy config.example.json to ~/.pve-mcp-server/config.json
(%USERPROFILE%\.pve-mcp-server\config.json on Windows) and fill in your values:
{
"host": "pve-node.domain.local",
"port": 8006,
"token_user": "user@realm",
"token_name": "tokenname",
"token_secret": "00000000-0000-0000-0000-000000000000"
}Recommended token role: PVEAuditor (read-only) — every tool issues GET requests only.
Related MCP server: Proxmox MCP Server
Use via uvx (no local install)
{
"mcpServers": {
"pve": {
"command": "uvx",
"args": ["--from", "git+https://github.com/UriWise/pve-mcp-server", "pve-mcp-server"]
}
}
}Local development
uv sync
uv run pve-mcp-serverTools
pve_list_vms, pve_get_vm_disks, pve_get_storage, pve_get_snapshots, pve_get_nodes,
pve_get_vm_config, pve_get_vm_status, pve_get_cluster_resources, pve_list_lxc,
pve_get_node_status, pve_get_node_network, pve_get_node_disks, pve_get_storage_content,
pve_get_cluster_tasks, pve_get_backup_jobs, pve_get_ha_status, pve_get_replication_jobs,
pve_get_firewall_rules, pve_get_vm_agent, pve_get_vm_rrddata, pve_get_node_rrddata,
pve_get_node_services, pve_get_node_pci, pve_get_vm_pending, pve_get_node_certificates
Available Tools
25 toolspve_get_backup_jobsA
List configured vzdump backup jobs: schedule, storage, VMs, compression, mode.
Inputs: no named parameters. Returns: result
pve_get_cluster_resourcesA
All cluster resources in one bulk call: VMs, nodes, storage, LXC. Fast overview.
Inputs: type. Returns: result
pve_get_cluster_tasksC
Cluster task history: migrations, backups, clones, snapshot operations.
Inputs: limit. Returns: result
pve_get_firewall_rulesB
Firewall rules for a node or a specific VM (pass vm_name for VM-level rules).
Inputs: node (required), vm_name. Returns: result
pve_get_ha_statusA
HA cluster status and list of HA-managed resources.
Inputs: no named parameters.
pve_get_node_certificatesC
TLS certificates on a node: subject, issuer, fingerprint, expiry dates.
Inputs: node (required). Returns: result
pve_get_node_disksA
Physical disks on a node: model, serial, size, type, S.M.A.R.T. health, wear.
Inputs: node (required). Returns: result
pve_get_node_networkB
All NICs and bridges on a node: IP, netmask, gateway, MAC, bridge ports.
Inputs: node (required). Returns: result
pve_get_node_pciC
PCI devices available for passthrough on a node (IOMMU groups, vendor, class).
Inputs: node (required). Returns: result
pve_get_node_rrddataC
Historical metrics for a node: CPU, RAM, net, disk over time.
Inputs: node (required), timeframe. Returns: result
pve_get_nodesA
List all nodes in the Proxmox cluster with status, memory, and uptime.
Inputs: no named parameters. Returns: result
pve_get_node_servicesA
systemd services on a node: pveproxy, pvedaemon, pve-ha-*, corosync, etc.
Inputs: node (required). Returns: result
pve_get_node_statusA
Detailed node metrics: CPU%, RAM, swap, load averages, disk, kernel, PVE version.
Inputs: node (required).
pve_get_replication_jobsA
List VM replication jobs: source, target, schedule, rate, status.
Inputs: no named parameters. Returns: result
pve_get_snapshotsA
List Proxmox-level snapshots for a VM (or all VMs if no name given).
Inputs: vm_name. Returns: result
pve_get_storageA
List storage pools with capacity info (type, total GB, free GB, used %).
Inputs: no named parameters. Returns: result
pve_get_storage_contentA
List content of a storage pool: VM images, ISOs, backups, templates.
Inputs: node (required), storage (required), content_type. Returns: result
pve_get_vm_agentA
QEMU guest agent data: IP addresses, OS info, hostname. Requires agent installed inside VM.
Inputs: vm_name (required).
pve_get_vm_configB
Full VM config: CPU type, BIOS, boot order, NIC list, disk config, tags, description.
Inputs: vm_name (required).
pve_get_vm_disksB
Get disk/storage layout for a specific VM: storage pool, path, size per disk.
Inputs: vm_name (required). Returns: result
pve_get_vm_pendingA
Config changes pending reboot for a VM (diff between live and stored config).
Inputs: vm_name (required). Returns: result
pve_get_vm_rrddataB
Historical metrics for a VM: CPU, RAM, net I/O, disk I/O over time.
Inputs: vm_name (required), timeframe. Returns: result
pve_get_vm_statusB
Live VM metrics: CPU%, RAM used/max, PID, uptime, QMP status, lock, tags.
Inputs: vm_name (required).
pve_list_lxcA
List LXC containers across all nodes (name, status, RAM, CPU, disk).
Inputs: node, status_filter. Returns: result
pve_list_vmsA
List QEMU VMs in the Proxmox cluster. Optionally filter by name (supports * wildcard) or status.
Inputs: name_filter, status_filter. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 25 tools
Every tool targets a distinct resource and aspect, such as node status vs. node network vs. node PCI devices. Even similar-sounding tools like pve_get_node_disks and pve_get_node_pci are clearly separated by described hardware type.
All tools share the pve_ prefix and use lowercase with underscores, but two tools use 'list_' (pve_list_vms, pve_list_lxc) while the rest use 'get_', creating a minor inconsistency in verb choice.
With 25 tools, the server sits at the heavy end of the borderline range. The count is reasonable for a comprehensive read-only monitoring surface, but it feels ample and lacks any action-oriented tools to balance it.
The toolset is entirely read-only, offering no ability to create, start, stop, delete, or modify virtual machines or containers. This leaves major management workflows completely unsupported, making it severely incomplete for a Proxmox VE server.
Maintenance
Related MCP Connectors
Read-only MCP access to a documented IT fleet: state, changes, posture. 15 tools.
Read-only developer, date, finance, and text utilities. Authless remote MCP server by Clean.tools.
2,000+ MCP servers read at source level. Know what one does before you connect. Free, no key.
Linux server administration over OAuth MCP: shell, Docker, systemd, files and monitoring.
Related MCP Servers
- AlicenseAqualityBmaintenanceMinimal MCP server for Proxmox VE — 38 tools for nodes, VMs, LXC, storage, and snapshots. Read-only by default, single Docker image, multi-arch.3881 PyPI7MIT
- AlicenseAqualityCmaintenanceMCP server for Proxmox VE and Proxmox Datacenter Manager, covering every API endpoint via six consolidated tools for list, describe, and call operations with a read-only safety gate.614 npm2AGPL 3.0
- AlicenseAqualityDmaintenanceA read-only MCP server for Proxmox VE that provides AI assistants with structured visibility into cluster nodes, guests, storage, and Docker workloads. It is designed to prevent any mutating operations by construction.12MIT
- FlicenseNot gradedqualityDmaintenanceProvides read-only Proxmox cluster oversight and configuration compliance checks through MCP, covering health, inventory, storage, backups, snapshots, and updates.-