The Proxmox MCP Server provides 92 tools for comprehensive management of Proxmox Virtual Environment, supporting both QEMU VMs and LXC containers.
Node Management
Query node info, status, logs, tasks, and network/DNS configuration
Manage node services (start/stop/restart), subscriptions, and APT packages
Control node power (shutdown/reboot/wake-on-LAN)
Perform bulk guest operations (start/stop/migrate all)
Manage node replication jobs, network interfaces (bridges, bonds, VLANs), and disk administration
VM & Container Lifecycle
List, create, start, stop, reboot, shutdown, pause/resume, clone, migrate, delete, and convert VMs/containers to templates
Update configuration, resize CPU/memory, manage disks, mount points (LXC), and network interfaces
Manage snapshots and backups (create, list, restore, delete)
QEMU Guest Agent
Query guest OS info, filesystem, network interfaces, time, hardware, hostname, and users
Execute commands, read/write files, manage filesystem freeze/thaw, control guest power, and set user passwords
Cluster & High Availability
Query/update cluster status and options; manage node joining
Manage HA resources and groups for automatic failover
Manage cluster backup and replication jobs
Firewall
Manage cluster-level and per-guest firewall rules, groups, aliases, and IP sets
Storage
Manage storage configurations (NFS, iSCSI, Ceph, LVM, ZFS, etc.)
List, upload, download, delete, and prune storage content; restore files from backups
SDN Networking
Manage SDN virtual networks (VNets), zones, controllers, and subnets
Access Control
Manage users, groups, roles, ACL entries, authentication domains (PAM, LDAP, AD, OpenID), user API tokens, and resource pools
Ceph
Manage Ceph cluster status, OSDs, monitors, MDS daemons, pools, and filesystems
Console Access
Get VNC, terminal, and SPICE proxy tickets for VMs and containers
Cloud-Init & Certificates
Manage cloud-init configurations for QEMU VMs
Manage node SSL certificates and ACME certificates/accounts
Notifications
Manage notification targets (SMTP, Gotify, Sendmail)
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., "@Proxmox MCP Server list all VMs and their status"
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.
Model Context Protocol (MCP) server for Proxmox Virtual Environment
English | 한국어
A comprehensive MCP server providing 92 tools for managing Proxmox Virtual Environment, including QEMU VMs and LXC containers.
This project is a TypeScript rewrite of mcp-proxmox-server, which itself was a Node.js port of the original Python implementation by canvrno/ProxmoxMCP.
Architecture:
3,147-line single-file monolith → modular TypeScript across 110+ source files
No type safety → strict TypeScript with noUncheckedIndexedAccess
Hand-written JSON Schema → Zod schemas with automatic JSON Schema generation
Giant switch statement (55 cases) → tool registry with handler/schema pairs
Quality:
0 tests → 1,134 tests
No input validation → Zod runtime validation on every tool call
Implicit error handling → structured MCP error responses with context
No permission checks → two-tier permission model (basic / elevated)
Developer Experience:
npx @bldg-7/proxmox-mcp just works
All 92 tool descriptions exposed via MCP ListTools
Rate limiter middleware included
Pino structured logging instead of console.log
Related MCP server: proxmox-mcp
92 comprehensive tools for Proxmox management
Full TypeScript implementation with strict type safety
Support for both QEMU VMs and LXC containers
Secure authentication (API token)
Flexible SSL modes (strict, verify, insecure)
Permission-based access control (basic vs elevated operations)
Comprehensive error handling with structured responses
Built on MCP SDK 1.25.3
npm install @bldg-7/proxmox-mcp
# or
pnpm add @bldg-7/proxmox-mcp
# or
yarn add @bldg-7/proxmox-mcpSet the following environment variables before starting the server:
Variable | Required | Description | Default |
| Yes | Proxmox server hostname or IP address | - |
| No | Username with realm (e.g., |
|
| Yes | API token name | - |
| Yes | API token value | - |
| No | SSL verification mode |
|
| No | Path to a CA certificate (PEM) trusted in | - |
| No | Allow elevated operations |
|
| No | Proxmox API port |
|
| No | Allow shell special characters in exec commands |
|
| No | Enable SSH-based LXC exec |
|
| No | SSH host (falls back to PROXMOX_HOST) | - |
| No | SSH port |
|
| No | SSH username |
|
| When SSH enabled | Path to SSH private key | - |
| When SSH enabled | Proxmox node name reachable via SSH | - |
| No | Host key fingerprint for verification | - |
| No | Log verbosity ( |
|
strict: Full SSL certificate verification against the system CA store (recommended for production)
verify: Same as strict, but additionally trusts the CA certificate given via PROXMOX_SSL_CA_CERT. For self-signed certificates, point PROXMOX_SSL_CA_CERT at the server certificate (or your internal CA) in PEM format. Without PROXMOX_SSL_CA_CERT, this mode behaves exactly like strict.
insecure: No SSL verification (development only, not recommended)
Do not use NODE_TLS_REJECT_UNAUTHORIZED=0 to work around certificate errors — it disables TLS verification for the entire process, and the server logs a warning when it is set. Use PROXMOX_SSL_MODE and PROXMOX_SSL_CA_CERT instead.
The server implements a two-tier permission model:
Basic operations: Read-only operations (list, get, status) - always allowed
Elevated operations: Create, modify, delete operations - require PROXMOX_ALLOW_ELEVATED=true
This prevents accidental destructive operations and provides an additional safety layer.
# With environment variables
export PROXMOX_HOST=pve.example.com
export PROXMOX_TOKEN_NAME=mytoken
export PROXMOX_TOKEN_VALUE=abc123-def456-ghi789
export PROXMOX_SSL_MODE=verify
export PROXMOX_SSL_CA_CERT=/path/to/proxmox-ca.pem # needed with verify for self-signed certs
export PROXMOX_ALLOW_ELEVATED=true
npx @bldg-7/proxmox-mcpAdd to your Claude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"proxmox": {
"command": "npx",
"args": ["-y", "@bldg-7/proxmox-mcp"],
"env": {
"PROXMOX_HOST": "pve.example.com",
"PROXMOX_TOKEN_NAME": "mytoken",
"PROXMOX_TOKEN_VALUE": "abc123-def456-ghi789",
"PROXMOX_SSL_MODE": "verify",
"PROXMOX_SSL_CA_CERT": "/path/to/proxmox-ca.pem",
"PROXMOX_ALLOW_ELEVATED": "true"
}
}
}
}This server provides 92 comprehensive tools for Proxmox management:
Category | Tools | Permission |
Node Management | 14 | Mixed |
Guest (VM/LXC) | 23 | Mixed |
Guest Creation | 2 | Elevated 🔒 |
QEMU Agent | 7 | Elevated 🔒 |
Cluster Operations | 5 | Mixed |
Cluster Firewall | 6 | Mixed |
High Availability | 2 | Mixed |
Storage | 3 | Mixed |
SDN Networking | 4 | Mixed |
Access Control | 7 | Mixed |
Ceph | 6 | Mixed |
Console Access | 3 | Elevated 🔒 |
Backup | 1 | Mixed |
VM Disks | 1 | Elevated 🔒 |
LXC Mount Points | 1 | Elevated 🔒 |
LXC Exec | 1 | Elevated 🔒 |
Cloud-Init | 1 | Mixed |
Certificates | 1 | Mixed |
ACME | 3 | Mixed |
Notifications | 1 | Mixed |
Total | 92 |
This package includes Agent Skills - AI-optimized documentation that teaches agents how to use Proxmox MCP tools and operational best practices. Skills follow the agentskills.io open standard and work across multiple AI agent platforms.
Skill | Description | Tools/Topics |
proxmox-mcp-tools | Complete MCP tool reference for Proxmox VE | 92 tools across 14 domains (VMs, LXC, cluster, storage, networking, Ceph, certificates, ACME, notifications) |
proxmox-admin | Operational expertise for Proxmox infrastructure | VM lifecycle, storage management, HA configuration, troubleshooting |
For Agent Skills Standard (Claude Code, OpenCode, Cursor, Codex, Gemini CLI, VS Code):
npx skills add Bldg-7/proxmox-mcpFor Claude Code Plugin:
/plugin marketplace add Bldg-7/proxmox-mcpClaude Code - Anthropic's official CLI
OpenCode - Open-source Claude Code alternative
Cursor - AI-powered code editor
Codex - GitHub Copilot CLI
Gemini CLI - Google's AI assistant
VS Code - With agent skills extensions
Skills provide progressive disclosure: metadata (~100 tokens) → instructions (<5000 tokens) → detailed references (on demand).
Once installed, skills are automatically loaded into the agent's context. The agent learns Proxmox tool usage, workflows, and best practices — so you can give natural language instructions and the agent translates them into the right MCP tool calls.
Example — Creating a VM (agent uses proxmox-mcp-tools skill):
You: "Create an Ubuntu VM with 4 cores, 8GB RAM, and 50GB disk on pve1"
Agent knows the workflow from the skill:
1. proxmox_get_next_vmid → gets available VM ID (e.g., 105)
2. proxmox_create_vm → creates VM 105 with 4 cores, 8GB RAM
3. proxmox_vm_disk → attaches 50GB virtio disk (action: 'add')
4. proxmox_guest_network → adds network interface on vmbr0 (action: 'add', type: 'vm')
5. proxmox_guest_start → powers on the VM (type: 'vm')Example — Setting up HA (agent uses proxmox-admin skill):
You: "Make VM 100 highly available"
Agent knows the operational playbook:
1. proxmox_ha_group → checks existing HA groups (action: 'list')
2. proxmox_ha_resource → adds VM 100 to HA with priority (action: 'create')
3. proxmox_ha_resource → verifies HA is active (action: 'status')Example — Troubleshooting (agent uses both skills together):
You: "VM 100 won't start, help me figure out why"
Agent combines tool knowledge + operational expertise:
1. proxmox_guest_status → checks current state (type: 'vm')
2. proxmox_guest_config → reviews configuration (action: 'get', type: 'vm')
3. proxmox_node → checks node resource availability (action: 'status')
4. proxmox_node_task → finds recent failed tasks (action: 'list')
→ Diagnoses: "Node pve1 has insufficient memory. VM requires 8GB but only 2GB free."
→ Suggests: resize VM memory, migrate to another node, or free resourcesproxmox-mcp-tools — Tool Reference:
92 tools organized into 14 domains (VMs, LXC, cluster, storage, networking, Ceph, access control, pools, certificates, ACME, notifications)
Parameters, types, and descriptions for every tool
Permission levels (basic vs elevated 🔒)
Common workflow patterns (create VM, backup/restore, clone, migrate)
proxmox-admin — Operational Expertise:
VM & LXC lifecycle playbooks (create → configure → monitor → backup → decommission)
Storage management strategies (Ceph, NFS, LVM, ZFS)
HA configuration and failover procedures
Troubleshooting guides for common API quirks
Security best practices (permission model, token management)
Performance monitoring and optimization
This package includes SubAgents - specialized AI agents for executing domain-specific Proxmox operations. While Skills provide knowledge ("here are the tools and how to use them"), SubAgents provide action ("I'll do this for you"). SubAgents automatically load the relevant skills and execute operations using MCP tools.
SubAgent | Domain | Description |
vm-manager | QEMU VMs | VM lifecycle management (create, configure, start/stop, snapshots, backups, cloning, templates) |
lxc-manager | LXC Containers | Container lifecycle management (create, configure, mount points, snapshots, backups, cloning) |
cluster-admin | Cluster Operations | High availability, VM/LXC migration, replication jobs, cluster firewall, cluster backup jobs |
storage-admin | Storage Infrastructure | Storage backends (NFS, iSCSI, Ceph), ISO/template management, backup pruning, disk operations |
network-admin | Network Infrastructure | SDN (VNets, zones, controllers, subnets), node network interfaces, DNS configuration |
access-admin | Access Control | Users, groups, roles, ACLs, authentication domains (PAM, LDAP, AD) |
monitor | Monitoring | Read-only health checks, status monitoring, log analysis, task tracking (delegates actions to other agents) |
SubAgents use auto-delegation in Claude Code and compatible platforms:
User makes request: "Create a new Ubuntu VM with 4 cores and 8GB RAM"
Claude analyzes request: Matches request to vm-manager agent based on example blocks
Auto-delegates: Invokes vm-manager agent with full context
Agent executes: vm-manager loads skills, calls MCP tools, completes operation
Returns result: Structured response with operation details
MCP Server: The Proxmox MCP server must be connected and configured
Skills: SubAgents automatically load proxmox-mcp-tools and proxmox-admin skills
Permissions: Elevated operations require PROXMOX_ALLOW_ELEVATED=true in MCP server config
SubAgents are included when you install the plugin:
For Claude Code Plugin:
/plugin marketplace add Bldg-7/proxmox-mcpFor Agent Skills Standard (if supported):
npx skills add Bldg-7/proxmox-mcpCreate a VM (auto-delegates to vm-manager):
User: "Create a new Ubuntu VM with 4 cores, 8GB RAM, and a 50GB disk on node pve1"
→ vm-manager executes: get next VMID → create VM → add disk → add network → startEnable HA (auto-delegates to cluster-admin):
User: "Enable high availability for VM 100 with priority 100"
→ cluster-admin executes: check HA config → add HA resource → verifyMonitor cluster (auto-delegates to monitor):
User: "Show me the current cluster status and any issues"
→ monitor executes: get cluster status → get HA status → get nodes → reportEach SubAgent has clear responsibilities to avoid overlap:
vm-manager: Only QEMU VMs (not LXC). Delegates cluster ops to cluster-admin.
lxc-manager: Only LXC containers (not VMs). Delegates cluster ops to cluster-admin.
cluster-admin: Cluster-wide operations (HA, firewall aliases/ipsets, cluster config, notifications). Delegates per-VM/LXC ops to respective managers.
storage-admin: Storage infrastructure, certificate management, ACME configuration, advanced disk operations. Delegates VM disk operations to vm-manager.
network-admin: Network infrastructure only. Delegates per-VM/LXC NICs to respective managers.
access-admin: Access control including API token management. No delegation to other agents.
monitor: Read-only monitoring including node metrics/RRD data. Delegates ALL actions to appropriate agents.
All tools return structured responses following the MCP protocol:
{
"content": [
{
"type": "text",
"text": "✅ Operation completed successfully\n\n• Details here..."
}
],
"isError": false
}{
"content": [
{
"type": "text",
"text": "❌ Error: Operation failed\n\nReason: Detailed error message"
}
],
"isError": true
}{
"content": [
{
"type": "text",
"text": "🚫 Permission Denied: Operation requires elevated permissions\n\nSet PROXMOX_ALLOW_ELEVATED=true to enable this operation."
}
],
"isError": true
}Problem: ECONNREFUSED or connection timeout
Solutions:
Verify PROXMOX_HOST is correct and reachable
Check PROXMOX_PORT (default: 8006)
Ensure firewall allows connections to Proxmox API port
Try PROXMOX_SSL_MODE=insecure for testing (not recommended for production)
Problem: 401 Unauthorized or authentication failed
Solutions:
Verify PROXMOX_USER includes realm if set (e.g., root@pam, not just root)
Verify PROXMOX_TOKEN_NAME and PROXMOX_TOKEN_VALUE are valid
Ensure API token has sufficient permissions in Proxmox
Problem: UNABLE_TO_VERIFY_LEAF_SIGNATURE or SSL errors
Solutions:
For self-signed certificates, set PROXMOX_SSL_MODE=verify and PROXMOX_SSL_CA_CERT=/path/to/cert.pem (the server certificate or your internal CA) — verify without PROXMOX_SSL_CA_CERT behaves like strict and still fails
Use PROXMOX_SSL_MODE=insecure for development (not recommended for production)
Install proper SSL certificates on Proxmox server for production use
Do not set NODE_TLS_REJECT_UNAUTHORIZED=0; it disables TLS verification process-wide
Problem: 🚫 Permission Denied for operations
Solutions:
Set PROXMOX_ALLOW_ELEVATED=true to enable elevated operations
Review which operations require elevated permissions (marked with 🔒)
Ensure this is intentional - elevated operations can modify/delete resources
git clone https://github.com/Bldg-7/proxmox-mcp.git
cd proxmox-mcp
pnpm install
pnpm buildpnpm test # Run tests once
pnpm test:watch # Run tests in watch mode
pnpm test:coverage # Run tests with coverage reportpnpm typecheck # Type check without emitting filespnpm lint # Run ESLintContributions are welcome! Please see CONTRIBUTING.md for guidelines.
MIT License - see LICENSE file for details.
Built with Model Context Protocol SDK
Powered by Proxmox Virtual Environment
Legend: 🔒 = Requires elevated permissions (PROXMOX_ALLOW_ELEVATED=true)
Manage ACL entries (get, update)
Inputs: action (required), path, roles, users, +7 more.
Manage ACME accounts. action=list|get: query accounts | action=create|update|delete: manage accounts (requires elevated)
Inputs: action (required), name, contact, tos_url, +1 more.
Manage ACME certificates. action=order|renew|revoke: certificate ops (requires elevated) | action=config: get ACME config
Inputs: node (required), action (required), force.
Query ACME information. action=list_plugins|get_plugin|directories
Inputs: action (required), id.
Execute commands via QEMU agent. operation=exec: run command | operation=status: check execution status (requires elevated permissions)
Inputs: node (required), vmid (required), operation (required), pid, +5 more.
Read/write files via QEMU agent. operation=read|write (requires elevated permissions)
Inputs: file (required), node (required), vmid (required), operation (required), +2 more.
Manage filesystem freeze via QEMU agent. operation=status|freeze|thaw|fstrim (requires elevated permissions)
Inputs: node (required), vmid (required), operation (required).
Query guest hardware via QEMU agent. operation=memory_blocks|vcpus|memory_block_info|hostname|users (requires elevated permissions)
Inputs: node (required), vmid (required), operation (required).
Query guest info via QEMU agent. operation=ping|osinfo|fsinfo|network_interfaces|time|timezone (requires elevated permissions)
Inputs: node (required), vmid (required), operation (required).
Guest power control via QEMU agent. operation=shutdown|suspend_disk|suspend_ram|suspend_hybrid (requires elevated permissions)
Inputs: node (required), vmid (required), operation (required).
Manage guest users via QEMU agent. operation=set_password (requires elevated permissions)
Inputs: node (required), vmid (required), password (required), username (required), +2 more.
Manage APT packages. action=update(elevated)|upgrade(elevated): package ops | action=versions: list versions
Inputs: node (required), action (required), package.
Manage guest backups (create, list, restore, delete) for VMs and LXC containers (requires elevated permissions)
Inputs: node (required), action (required), mode, type, +5 more.
Query Ceph cluster. action=status: get Ceph cluster health, FSID, monitors, OSDs, and placement groups
Inputs: node (required), action (required).
Manage Ceph filesystems (list, create)
Inputs: node (required), action (required), name, pool, +2 more.
Manage Ceph MDS daemons (list, create, delete)
Inputs: node (required), action (required), name.
Manage Ceph monitors (list, create, delete)
Inputs: node (required), action (required), monid.
Manage Ceph OSDs (list, create, delete)
Inputs: node (required), action (required), id, dev, +5 more.
Manage Ceph pools (list, create, update, delete)
Inputs: node (required), action (required), name, size, +4 more.
Manage node SSL certificates. action=list: view certs | action=upload: upload custom cert (requires elevated) | action=delete: remove custom cert (requires elevated)
Inputs: node (required), action (required), key, force, +2 more.
Manage cloud-init for a QEMU VM. action=get: list config | action=dump: dump rendered config (dump_type=user|network|meta) | action=regenerate: regenerate drive (requires elevated)
Inputs: node (required), vmid (required), action (required), dump_type.
Query Proxmox cluster info. action=status: overall cluster status with nodes and resource usage | action=options: get cluster-wide options | action=update_options: update cluster-wide options (requires elevated permissions)
Inputs: action (required), options.
Manage cluster backup jobs. action=list: list jobs | action=get: get job by ID | action=create: create job (elevated) | action=update: update job (elevated) | action=delete: delete job (elevated)
Inputs: action (required), id, all, dow, +34 more.
Manage cluster config. action=get: get config | action=list_nodes: list config nodes | action=get_node: get node config | action=join: join cluster (elevated) | action=totem: get totem config
Inputs: action (required), node, force, hostname, +2 more.
Query/manage cluster firewall metadata. action=get_options: get firewall options | action=update_options: update firewall options (elevated) | action=list_macros: list firewall macros | action=list_refs: list firewall refs
Inputs: action (required), type, enable, policy_in, +2 more.
Manage cluster firewall aliases. action=list: list aliases | action=get: get alias by name | action=create: create alias (elevated) | action=update: update alias (elevated) | action=delete: delete alias (elevated)
Inputs: action (required), cidr, name, rename, +1 more.
Manage cluster firewall groups. action=list: list groups | action=get: get group by name | action=create: create group (elevated) | action=update: update group (elevated) | action=delete: delete group (elevated)
Inputs: action (required), group, delete, digest, +2 more.
Manage cluster firewall IP sets. action=list: list IP sets | action=create: create IP set (elevated) | action=delete: delete IP set (elevated)
Inputs: action (required), name, comment.
Manage cluster firewall IP set entries. action=list: list entries | action=create: add entry (elevated) | action=update: update entry (elevated) | action=delete: delete entry (elevated)
Inputs: name (required), action (required), cidr, comment, +1 more.
Manage cluster firewall rules. action=list: list rules | action=get: get rule by position | action=create: create rule (elevated) | action=update: update rule (elevated) | action=delete: delete rule (elevated)
Inputs: action (required), log, pos, dest, +13 more.
Manage cluster replication jobs. action=list: list jobs | action=get: get job by ID | action=create: create job (elevated) | action=update: update job (elevated) | action=delete: delete job (elevated)
Inputs: action (required), id, keep, rate, +10 more.
Get a SPICE proxy ticket for a QEMU VM (requires elevated permissions)
Inputs: node (required), vmid (required).
Get a terminal proxy ticket for a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Get a VNC proxy ticket for a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Create a new LXC container (requires elevated permissions)
Inputs: node (required), vmid (required), ostemplate (required), net0, +5 more.
Create a new QEMU virtual machine (requires elevated permissions)
Inputs: node (required), vmid (required), iso, name, +7 more.
Manage authentication domains (list, get, create, update, delete)
Inputs: action (required), port, type, realm, +14 more.
Restore files from backup. action=list: list files in backup | action=download: download file from backup
Inputs: node (required), action (required), volume (required), storage (required), +2 more.
Get the next available VM/Container ID number
Inputs: no named parameters.
Manage Proxmox groups (list, create, update, delete)
Inputs: action (required), users, append, delete, +3 more.
Clone a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required), newid (required), +2 more.
Get hardware configuration for a VM (type=vm) or LXC container (type=lxc)
Inputs: node (required), type (required), vmid (required).
Update VM/LXC config key-value pairs (type=vm|lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required), config, +1 more.
Delete a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Move guest storage for VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: disk (required), node (required), type (required), vmid (required), +2 more.
Resize guest storage for VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: disk (required), node (required), size (required), type (required), +1 more.
Check if a feature (snapshot, clone, copy) is available for a VM (type=vm) or LXC container (type=lxc)
Inputs: node (required), type (required), vmid (required), feature (required).
Manage per-guest firewall rules. action=list|get: query rules | action=create|update|delete: manage rules (elevated). type=vm|lxc. Uses rule_action for firewall action (ACCEPT/REJECT/DROP), rule_type for direction (in/out/group).
Inputs: node (required), type (required), vmid (required), action (required), +16 more.
List all virtual machines and containers across the cluster with their status
Inputs: node, type.
Migrate a VM or LXC container to another node (type=vm|lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required), target (required), +5 more.
Manage guest network interfaces. action=add|update|remove with type=vm|lxc (requires elevated permissions)
Inputs: net (required), node (required), type (required), vmid (required), +8 more.
Pause a QEMU virtual machine — VM only (requires elevated permissions)
Inputs: node (required), vmid (required).
Get pending configuration changes for a VM (type=vm) or LXC container (type=lxc)
Inputs: node (required), type (required), vmid (required).
Reboot a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Resize VM/LXC CPU or memory (type=vm|lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required), cores, +1 more.
Resume a paused QEMU virtual machine — VM only (requires elevated permissions)
Inputs: node (required), vmid (required).
Get performance metrics (RRD data) for a VM (type=vm) or LXC container (type=lxc)
Inputs: node (required), type (required), vmid (required), cf, +1 more.
Gracefully shutdown a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Manage guest snapshots (create, list, rollback, delete) for VMs and LXC containers (requires elevated permissions)
Inputs: node (required), type (required), vmid (required), action (required), +2 more.
Start a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Get detailed status for a VM (type=vm) or LXC container (type=lxc)
Inputs: node (required), type (required), vmid (required).
Forcefully stop a VM (type=vm) or LXC container (type=lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Convert a VM or LXC container to a template (type=vm|lxc) (requires elevated permissions)
Inputs: node (required), type (required), vmid (required).
Manage HA groups. action=list: list groups | action=get: get group details | action=create: create group (elevated) | action=update: update group (elevated) | action=delete: delete group (elevated)
Inputs: action (required), type, group, nodes, +5 more.
Manage HA resources. action=list: list resources | action=get: get resource details | action=status: get HA manager status | action=create: create resource (elevated) | action=update: update resource (elevated) | action=delete: delete resource (elevated)
Inputs: action (required), sid, type, group, +7 more.
Execute a command inside an LXC container via SSH + pct exec (requires elevated + PROXMOX_SSH_ENABLED=true)
Inputs: node (required), vmid (required), command (required), timeout.
Manage LXC mount points. action=add: add mountpoint | action=remove: remove mountpoint (all elevated)
Inputs: mp (required), node (required), vmid (required), action (required), +2 more.
Query Proxmox node info. action=list: list all nodes | action=status: node status (elevated) | action=network: network interfaces | action=dns: DNS config | action=iface: specific interface details
Inputs: action (required), node, type, iface.
Bulk guest operations. action=start_all|stop_all|migrate_all (all elevated)
Inputs: node (required), action (required), target, maxworkers, +1 more.
Manage node configuration. action=get_time|set_time(elevated)|set_dns(elevated)|get_hosts|set_hosts(elevated)
Inputs: node (required), action (required), ip, dns1, +9 more.
Query node disk information. action=list|smart|lvm|zfs|lvmthin|directory
Inputs: node (required), action (required), disk, type, +3 more.
Destructive disk operations. action=init_gpt: initialize GPT | action=wipe: wipe disk (all elevated)
Inputs: disk (required), node (required), action (required), uuid.
Query node information. action=aplinfo: appliance templates | action=netstat: network stats | action=rrddata: performance metrics | action=storage_rrddata: storage metrics | action=report: diagnostic report
Inputs: node (required), action (required), cf, storage, +1 more.
Read node logs. action=syslog: read syslog | action=journal: read systemd journal
Inputs: node (required), action (required).
Manage node network interfaces. action=create|update|delete|apply (all elevated)
Inputs: node (required), action (required), mtu, cidr, +21 more.
Node power control. action=shutdown|reboot|wakeonlan (all elevated)
Inputs: node (required), action (required).
Manage node replication. action=status|log: query | action=schedule: trigger now (elevated)
Inputs: id (required), node (required), action (required).
Manage node services. action=list: list services | action=control: start/stop/restart a service (elevated)
Inputs: node (required), action (required), command, service.
Manage node subscription. action=get: get info | action=set: set key (elevated) | action=delete: remove (elevated)
Inputs: node (required), action (required), key.
Query node tasks. action=list: list recent tasks | action=get: get task details by UPID
Inputs: node (required), action (required), upid.
Manage notification targets. action=list|get: query targets | action=create|delete|test: manage targets (requires elevated)
Inputs: action (required), from, mode, name, +11 more.
Manage resource pools (list, get, create, update, delete)
Inputs: action (required), delete, digest, poolid, +1 more.
Manage Proxmox roles (list, create, update, delete)
Inputs: action (required), privs, append, delete, +3 more.
Manage SDN controllers (list, get, create, update, delete)
Inputs: action (required), ip, port, type, +7 more.
Manage SDN subnets (list, get, create, update, delete)
Inputs: action (required), dns, mtu, cidr, +9 more.
Manage SDN virtual networks (list, get, create, update, delete)
Inputs: action (required), mac, mtu, tag, +10 more.
Manage SDN zones (list, get, create, update, delete)
Inputs: action (required), mtu, tag, ipam, +8 more.
Manage storage configurations (list, get, create, update, delete, cluster_usage)
Inputs: action (required), node, path, pool, +25 more.
Manage storage content (list, list_templates, upload, download_url, delete, prune)
Inputs: node (required), action (required), storage (required), url, +16 more.
Manage Proxmox users (list, get, create, update, delete)
Inputs: action (required), email, append, delete, +9 more.
Manage user API tokens (list, get, create, update, delete)
Inputs: action (required), userid (required), expire, comment, +2 more.
Manage VM disks. action=add: add disk | action=remove: remove disk (all elevated)
Inputs: disk (required), node (required), vmid (required), action (required), +2 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 92 tools
Multiple tool boundaries are unclear: proxmox_guest_resize (CPU/memory) vs proxmox_guest_disk_resize (storage) are easily confused, and node queries are split across proxmox_node, proxmox_node_info, and proxmox_node_config with overlapping network/DNS actions. Disk operations are also scattered across proxmox_vm_disk and proxmox_guest_disk_*, creating ambiguous choices.
Most tools follow a proxmox_<resource>_<action> snake_case pattern, but the set mixes action-in-tool-name (proxmox_guest_start) with action-parameter designs (proxmox_node, proxmox_cluster). There are also outliers like proxmox_create_vm/proxmox_create_lxc (versus the generic proxmox_guest_* Family) and proxmox_get_next_vmid which breaks the resource-first convention.
With 92 tools, this server is far beyond an effective scope and will overwhelm agent tool selection. Even for a full hypervisor-management API, grouping related operations would make the set far more usable.
The tool surface is impressively broad and covers most core Proxmox domains: guest lifecycle (creation, start/stop, clone, migrate, template), snapshots, backups, storage, Ceph, SDN, users/ACLs, firewall, HA, and node administration. Minor gaps exist (e.g., no explicit guest rename, backup restore lacks granular options), but common workflows are essentially fully covered.
MCP server with quote and live cryptocurrency price tools, local and cloud-deployed transports.
A MCP server built for developers enabling Git based project management with project and personal…
MCP registry & directory: search, find & install 31k+ MCP servers & tools. Catalog and marketplace.