Skill: ops-proxmox

Inline

Proxmox VE infrastructure with Terraform (VMs, LXC, network, storage, backup)

Configuration

Property	Value
Context	inline
Pre-approved tools	None declared: the project's permission settings apply
Trigger	keywords: ops, proxmox, pve, proxmox ve, proxmox vm, proxmox lxc

Detailed description

Proxmox Infrastructure Skill

Proxmox VE infrastructure management with Terraform: provisioning of virtual machines, LXC containers, network configuration, storage and backup.

When to use this skill

This skill is automatically activated when the conversation mentions:

• "Proxmox", "PVE", "Proxmox VE"
• "Proxmox VM", "Proxmox LXC", "Proxmox container"
• "Proxmox cluster", "Proxmox node"
• "PBS", "Proxmox Backup Server"
• "Proxmox cloud-init"
• "QEMU/KVM" in a Proxmox context

Core principles

1. Infrastructure as Code

All Proxmox infrastructure must be managed via Terraform:

• Reproducibility: same config = same result
• Versioning: history in Git
• Review: PR to validate infra changes
• Documentation: the code IS the documentation

2. Environment separation

environments/
├── dev/           # Can be destroyed
├── staging/       # Mirrors prod
└── prod/          # Critical

Each environment has its own variables (terraform.tfvars), its own Terraform state and its own credentials.

3. Reusable modules

modules/
├── vm/            # QEMU/KVM virtual machine
├── lxc/           # LXC container
├── network/       # Network configuration
├── storage/       # Storage configuration
└── backup/        # PBS configuration

Proxmox architecture

Resource hierarchy

Datacenter
├── Cluster (optional)
│   ├── Node 1 (pve1) → VMs, LXC, Storage, Network
│   ├── Node 2 (pve2)
│   └── Node 3 (pve3)
├── Storage (datacenter level)
│   ├── local, local-lvm (per node)
│   ├── nfs-shared (shared)
│   └── ceph (distributed)
└── SDN (Zones, VNets, Subnets)

Resource types

Type	Description	Use case
VM (QEMU)	Full virtual machine	Heavy workloads, strong isolation
LXC	System container	Lightweight services, high density
Template	Base image	Fast cloning of VMs/LXC
Snippet	cloud-init files	Automated configuration

Terraform provider

bpg/proxmox (recommended)

Modern, well-maintained provider, full coverage of the Proxmox API.

terraform {
  required_version = ">= 1.5.0"
  required_providers {
    proxmox = {
      source  = "bpg/proxmox"
      version = "~> 0.50"
    }
  }
}

provider "proxmox" {
  endpoint  = var.proxmox_endpoint
  api_token = var.proxmox_api_token  # Recommended token
  insecure  = var.proxmox_insecure   # Dev only

  ssh {
    agent    = true
    username = "root"
  }
}

API token authentication

# On the Proxmox node
pveum user token add terraform@pve terraform-token --privsep=0

# Minimal permissions
pveum aclmod / -user terraform@pve -role PVEVMAdmin
pveum aclmod /storage -user terraform@pve -role PVEDatastoreUser

Format: terraform@pve!terraform-token=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx

Naming conventions

VMs and containers

Environment	Pattern	Example
Production	prod-{role}-{index}	prod-web-01
Staging	stg-{role}-{index}	stg-api-01
Development	dev-{role}-{index}	dev-db-01
Test	test-{purpose}	test-migration

VMID ranges

Range	Usage
100-199	Infrastructure (DNS, DHCP, etc.)
200-299	Production
300-399	Staging
400-499	Development
500-599	Test/Temporary
9000-9099	Templates

Recommended tags

environment:prod
role:webserver
team:platform
backup:daily
managed-by:terraform
criticality:high

Security

Best practices

1. API Token: minimal permissions (dedicated role, not root)
2. Firewall: enable the Proxmox firewall by default
3. Isolation: separate VLANs per environment
4. Unprivileged LXC: always use unprivileged containers
5. Audit: log API and SSH access
6. Secrets: NEVER hardcode in HCL (use Vault, TF_VAR_*, or gitignored tfvars)

Minimal Terraform permissions

# Dedicated role
pveum role add TerraformRole -privs "VM.Allocate VM.Clone VM.Config.CDROM VM.Config.CPU VM.Config.Cloudinit VM.Config.Disk VM.Config.HWType VM.Config.Memory VM.Config.Network VM.Config.Options VM.Monitor VM.Audit VM.PowerMgmt Datastore.AllocateSpace Datastore.AllocateTemplate Datastore.Audit SDN.Use"

# User + assignment
pveum user add terraform@pve
pveum aclmod / -user terraform@pve -role TerraformRole

References

This SKILL.md section contains the core principles. For technical details with full HCL examples, see the reference files:

File	Content
references/terraform-modules.md	VM modules, LXC, usage, network, storage
references/cloud-init.md	cloud-config templates, snippet uploads
references/backup-ha.md	PBS schedule, commands, HA configuration
references/troubleshooting.md	Common issues, diagnostic commands, recovery

Rules

IMPORTANT: NEVER manage Proxmox manually via the UI. Always via Terraform.

IMPORTANT: Use unprivileged LXC by default (limited privilege escalation).

IMPORTANT: One Terraform state per environment (dev/staging/prod isolated).

YOU MUST use the bpg/proxmox provider (modern, maintained) rather than telmate/proxmox (deprecated).

YOU MUST use API tokens with minimal permissions, never root.

NEVER hardcode secrets in committed HCL. Use gitignored tfvars or Vault.

NEVER skip PBS backups on critical VMs.

Attribution

This skill is based on:

• Official Proxmox VE documentation
• bpg/proxmox Terraform provider

Automatic triggering

This skill is automatically activated when:

• The matching keywords are detected in the conversation
• The task context matches the skill's domain

Triggering examples

• "I want to ops..."
• "I want to proxmox..."
• "I want to pve..."

Context inline

Inline means the skill runs in the conversation (the default, no context key):

• Sees the full history
• Its work is visible as it happens
• Right for almost every skill

---

See also

• Back to skills
• Architecture