AI assistants

An AI assistant that can run shell commands can use cv4pve-autosnap to tell you which snapshots the tool has taken, and to take snapshots or apply a retention when you ask. Unlike cv4pve-report and cv4pve-diag, snap and clean change the cluster: they create and delete snapshots.

The cv4pve-autosnap skill

The repository ships a skill, skills/cv4pve-autosnap/SKILL.md: a short file in the Agent Skills format that Claude Code, Codex and other assistants load when a request matches its description. It tells the assistant to:

• connect only with the options file, passed with @, and never print it;
• run status freely: it only reads;
• before any snap or clean, run the same command with --dry-run, show you what it would create and remove, and run it only after you agree.

Install it with the skills command, which needs Node.js:

npx skills add Corsinvest/cv4pve-autosnap

Or copy the file into the skills folder of your assistant: .claude/skills/ for Claude Code,.agents/skills/ for Codex, in the project or under your home folder:

mkdir -p .claude/skills/cv4pve-autosnap
curl -fL -o .claude/skills/cv4pve-autosnap/SKILL.md https://raw.githubusercontent.com/Corsinvest/cv4pve-autosnap/master/skills/cv4pve-autosnap/SKILL.md

Give it a token of its own

The skill tells the assistant what to do; the token decides what it can do. The assistant can read the token and call the Proxmox VE API without cv4pve-autosnap, and the skill cannot prevent that. VM.Snapshot also allows rolling a guest back to any of its snapshots: grant it, with VM.Audit, only on the guests or the pool the assistant may snapshot, not on /vms: see Permissions. For questions only, VM.Audit is enough. Put the connection in an options file, so the token stays out of the conversation.