vchaindz

Claude Vsphere Skill — Development skill for Claude Code

Development community

A Claude Code skill that turns Claude into a vSphere administrator using govc.

How to install Claude Vsphere Skill

This entry records only its repository, not the path inside it, so there is no exact command to give. Open vchaindz/claude-vsphere-skill and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Claude Vsphere Skill does

A Claude Code skill that turns Claude into a vSphere administrator using govc. Inventory, reports, snapshot audits, root-cause analysis, and safe VM/host/cluster management for VMware vCenter and ESXi — read-only first, with confirmation guardrails for destructive operations.

Alternatives in Development

Root Cause Tracing — Use when errors occur deep in execution and you need to trace back to find the original trigger 102.5k ★
Om Tidy — Self-maintenance pass — acts on every hygiene flag: archives completed work, groups loose clusters, splits ove 4.6k ★
Analyze Cluster Postmortem — Postmortem analysis of cluster failures → OPTIMIZE PROMPTS to work for EVERY use case 1.7k ★

README

claude-vsphere-skill

A [Claude Code](https://docs.claude.com/en/docs/claude-code) skill that turns Claude into a vSphere administrator using [govc](https://github.com/vmware/govmomi/tree/main/govc) — inventory, reporting, snapshot audits, root-cause analysis, and safe VM/host/cluster management for VMware vCenter and ESXi.

Ask in plain language:

"Which VMs have snapshots, and how old are they?" "VM web-01 was down last night. Find out what happened and when." "Datastore capacity report — flag anything over 85% full." "Prepare esx02 for patching." "Health check as an HTML file I can send to the customer." ([what that looks like](#reports-you-can-hand-to-someone))

And Claude picks the right govc commands, batches queries efficiently, parses the JSON, and answers like a colleague — including pushing back when your premise is wrong:

![Claude disproving an outage claim using vmware.log forensics](images/claude4.png)

*(Real session against a real vCenter: asked about an outage "last night", Claude proved from rotated `vmware.log` files — beyond vCenter's event retention — that the VM had been down since December, and reconstructed the host outage that caused it. All read-only.)*

Three requests, end to end

What the skill is actually for, in the words you would use.

**1. The morning report** — *"Run a health check on the vSphere environment and write it to `~/reports/` as HTML."*

Nine fixed checks in a fixed order — hosts, alarms, datastore capacity, snapshots, VMware Tools, cluster HA/DRS, orphaned VMs, recent errors, and an opt-in orphaned-VMDK scan. Same checks every run, so today's report is comparable with last week's, and a baseline file next to it turns the second run into *"what changed since yesterday"*. You get a self-contained HTML file, severity-ranked, with a machine-readable summary line so a cron wrapper can tell success from silence.

Read-only throughout: runs at tier `readonly` and works unattended. See [`docs/scheduled-report