VMware AIops

VMware AIops

For a read-only check of this skill with your model and agent host, use the shared model-fit self-test in the marketplace root.

Author: Wei Zhou, VMware by Broadcom — [email protected] This is a community-driven project by a VMware engineer, not an official VMware product. For official VMware developer tools see developer.broadcom.com.

English | 中文

AI-powered VMware vCenter/ESXi VM lifecycle and deployment tool — 61 tools.

Companion skills handle everything else:

Skill	Scope	Install
vmware-monitor	Read-only: inventory, health, alarms, events, metrics	uv tool install vmware-monitor
vmware-storage	Datastores, iSCSI, vSAN management	uv tool install vmware-storage
vmware-vks	Tanzu Namespaces, TKC cluster lifecycle	uv tool install vmware-vks

Need read-only monitoring only? Use VMware-Monitor — zero destructive code in the codebase.

[ClawHub] [Skills.sh] [Claude Code Marketplace] [License: MIT]

Upgrading to 2.0

2.0 renames 38 of its MCP tools so every name starts with a verb (e.g. cluster_add_host → add_cluster_host, cluster_configure → configure_cluster, cluster_create → create_cluster). The full old → new table is in RELEASE_NOTES.md. Nothing else changes: same tools, parameters, safety gates and CLI commands. If an agent prompt, script or ~/.vmware/rules.yaml names a tool, update it (but not the action words in saved plan files — cluster_add_host, storage_rescan and the like are plan actions, not tool names, and keep their spelling) — though deny rules written with the old names keep working with vmware-policy ≥ 1.18 (vmware-audit policy lists any rule the renames widened), and vmware-pilot ≥ 1.13 translates old names in saved workflows. If an install older than 2.0 shares that rules.yaml, give it vmware-policy ≥ 1.18.2 too: it matches each rule against both the old and the new name, so one file serves both.

⚡ Quick Investigation Reports (read-only)

Triage → investigate → act, all in one conversation. Five opinionated read-only reports aggregate and correlate server-side and hand back a high-signal result (never raw inventory), so you can decide where to look before changing anything. Each renders a self-contained offline HTML snapshot with --html (no external assets; drill-down detail collapses in native <details>, zero JavaScript). All delegate to the vmware-monitor library using AIops's own vCenter connection.

vmware-aiops attention                            # what needs attention now, all vCentersvmware-aiops investigate vm web-01 --hours 72     # everything around a VM, then act on itvmware-aiops investigate vm web-01 --html         # → offline snapshot in ~/vmware-health/

Via MCP these are the tools get_cluster_health_summary, get_cross_vcenter_attention, investigate_vm, investigate_host, investigate_datastore. (Requires vmware-monitor installed.)

Quick Install (Recommended)

Works with Claude Code, Cursor, Codex, Gemini CLI, Trae, and 30+ AI agents:

# Via Skills.shnpx skills add vmware-skills/VMware-AIops
# Via ClawHubclawhub install @zw008/vmware-aiops

PyPI Install (No GitHub Access Required)

# Install via uv (recommended)uv tool install vmware-aiops
# Or via pippip install vmware-aiops
# China mainland mirror (faster)pip install vmware-aiops -i https://pypi.tuna.tsinghua.edu.cn/simple

Offline / Air-Gapped Install (from source)

This project uses the modern PEP 517 build system (hatchling), so there is no setup.py by design — that is expected, not a missing file. If you cloned the source and hit ERROR: File "setup.py" or "setup.cfg" not found ... editable mode currently requires a setuptools-based build, your pip is older than 21.3 and cannot do an editable (-e) install with a non-setuptools backend. Editable mode is a developer convenience, not needed to run the tool — do one of:

# From the source tree — a normal (non-editable) install builds a wheel:pip install .              # NOT  pip install -e .
# ...or upgrade pip first, and editable works too:pip install --upgrade pip && pip install -e .

For a truly air-gapped host, build the wheels on a connected machine and copy them over — the target then needs no network:

# On a connected machine, collect this package + its dependencies as wheels:pip wheel . -w dist        # → dist/*.whl   (or: uv build, for just this package)
# Copy dist/ to the air-gapped host, then install offline:pip install --no-index --find-links dist vmware-aiops

Why this over other VMware MCP servers

Most open-source VMware MCP servers (e.g. bright8192/esxi-mcp-server, giuliolibrando/vmware-vsphere-mcp-server) are single-vCenter VM wrappers: list/power/snapshot a VM, basic monitoring, a confirm=True flag. They explicitly do not cover networking, storage, Kubernetes, ops analytics, load balancing, or compliance — and "logging is documented" is not an audit trail.

This is one skill in an 11-package family that covers the whole estate and runs every tool through a governed harness:

If you only ever power-cycle VMs in one vCenter, a single-file server is fine. If you run a real (regulated, NSX-segmented, multi-domain) VMware estate and need an AI operator an auditor can sign off on, that's what this family is for — see docs/compliance-ready.md.

Capabilities Overview

What This Skill Does

CLI vs MCP: Which Mode to Use

Rule of thumb: Use CLI for cost efficiency and small models. Use MCP for structured automation with large models.

Architecture

User (Natural Language)  ↓AI CLI Tool (Claude Code / Gemini / Codex / Aider / Continue / Trae / Kimi)  ↓ reads SKILL.md / AGENTS.md / rules  ↓vmware-aiops CLI  ↓ pyVmomi (vSphere SOAP API)  ↓vCenter Server ──→ ESXi Cluster ──→ VM    orESXi Standalone Host ──→ VM

Version Compatibility

pyVmomi auto-negotiates the API version during SOAP handshake — no manual configuration needed. The same codebase manages 7.0 / 8.0 / 9.0 / 9.1 environments seamlessly.

Official Broadcom References

• SDKs: https://developer.broadcom.com/sdks — VCF Python SDK (recommended for VCF 9+, bundles pyVmomi + vSAN SDK), vSphere Automation SDK for Python (REST APIs)
• REST APIs: https://developer.broadcom.com/xapis — vSphere Automation API, VCF API, SDDC Manager API
• CLI Tools: https://developer.broadcom.com/tools — PowerCLI 9.1, ESXCLI, OVF Tool

Common Workflows

Deploy a Lab Environment

1. Browse datastore for OVA images → vmware-aiops datastore browse <ds> --pattern "*.ova"
2. Deploy VM from OVA → vmware-aiops deploy ova ./image.ova --name lab-vm --datastore ds1
3. Install software inside VM → vmware-aiops vm guest-exec lab-vm --cmd /bin/bash --args "-c 'apt-get install -y nginx'" --user root
4. Create baseline snapshot → vmware-aiops vm snapshot-create lab-vm --name baseline
5. Set TTL for auto-cleanup → vmware-aiops vm set-ttl lab-vm --minutes 480

Batch Clone for Testing

1. Create plan: create_vm_plan with multiple clone + reconfigure steps
2. Review plan with user (shows affected VMs, irreversible warnings)
3. Apply: apply_vm_plan executes sequentially, stops on failure
4. If failed: rollback_vm_plan reverses executed steps
5. Set TTL on all clones for auto-cleanup

Migrate VM to Another Host

1. Check VM info via vmware-monitor → verify power state and current host
2. Migrate: vmware-aiops vm migrate my-vm --to-host esxi-02
3. Verify migration completed

VM Lifecycle

Guest Operations require VMware Tools running inside the guest OS, and the guest account is always named explicitly — --user on the CLI, username over MCP. There is no default, so no call runs as root without choosing root. run_vm_guest_command_with_output (MCP only; the CLI has no equivalent) auto-detects Linux/Windows shell and captures stdout/stderr.

Plan → Apply (Multi-step Operations)

For complex operations involving 2+ steps or 2+ VMs, use the plan/apply workflow instead of executing individually:

Plans stored in ~/.vmware-aiops/plans/, auto-deleted on success, auto-cleaned after 24h.

VM Deployment & Provisioning

Cluster Management

remove-host requires the host to be in maintenance mode first; the host is moved out of the cluster into the datacenter's host folder as a standalone host.

Alarm Management

Blast radius: vSphere has no per-alarm clear API. alarm reset uses AlarmManager.ClearTriggeredAlarms, which clears all triggered alarms matching the named alarm's entity type (host/VM/all) and current status (red/yellow) — not just the named one. The named alarm is looked up first (typos fail fast), and the output's scope field reports exactly what was cleared. Cleared alarms re-trigger automatically if their underlying condition persists.

Datastore Browser

Scheduled Scanning & Notifications

Safety Features

vCenter vs ESXi Comparison

Inventory, alarms, events, sensors, host services, and scanning are now in vmware-monitor.

What protects you

The table above lists two different surfaces and it is worth being blunt about which protections apply to which, because getting this wrong is worse than having no protection at all — a guardrail you believe in is one you stop compensating for.

On the CLI, a destructive command asks twice and takes no bypass flag, and --dry-run previews every write except deploy iso, deploy mark-template, vm cancel-ttl and vm guest-download. That defends a mistyped command typed by a human. It does not defend against an agent, which satisfies both prompts with yes |.

Over MCP, every tool annotated destructive, plus migrate_vm and the network/DRS authoring tools — 23 of the 43 write tools, including power_off_vm, migrate_vm, delete_cluster, the snapshot, guest, TTL, Clean Slate and plan tools, and reset_vcenter_alarm — takes one argument, confirm, whose default is a no-write preview: a bare call returns its blast radius and changes nothing. confirm=True re-measures and is refused, with a teaching error audited as a failure, if the preview found a blocker (a VM without running VMware Tools, a target host in maintenance mode, a cluster that still has hosts, a missing or duplicated snapshot) or could not read something. delete_vm also takes the preview's acknowledge_with echoed back, and is refused if the VM changed since, is powered on or suspended. The other 20 write tools — creates, clones, deploys, power-on, reconfigure, snapshot create, host add, template and ISO operations, guest download, plan creation, alarm acknowledge — act on the first call. A confirmation is not authorization, which is why the VMWARE_READ_ONLY switch stays removed (it was enforced on the MCP path only, and any agent with a shell walked around it via the CLI). What the preview buys is narrower: an agent does not destroy something it has not looked at.

What actually decides whether a write lands is the vCenter/ESXi service account. Give the skill an account with the privileges the work needs and no more; vCenter refuses the rest itself, on every surface, with no way around it from inside this skill. To run an agent read-only, give it a read-only vCenter role — one decision, enforced where it is made. Every call is then recorded in ~/.vmware/audit.db before the caller sees a result, which is how you find out what happened. Optional deny rules in ~/.vmware/rules.yaml, checked before every MCP call, can refuse operations — for example, writes to targets labelled environment: production. The shipped baseline denies nothing, and the rules run inside the same process: a guardrail on top of RBAC, not a replacement.

run_vm_guest_command is the one to think hardest about. It runs a caller-supplied command inside the guest OS with the credentials handed to it — its username is required (no default account); nothing bounds what the command may be. The guest account is a separate authorization boundary from the vCenter one — a read-only vCenter role does not constrain what this tool does inside a VM. The skill stores no guest credentials: over MCP they are tool arguments the agent sees (the audit row redacts the password). Pass a least-privilege guest account, and do not hand the agent guest credentials it does not need. upload_vm_guest_file reads any local file the server process can read.

The full inventory of which tools are gated and which are not is in references/capabilities.md, where the numbers are checked against the live tool registry by the test suite rather than maintained by hand.

Troubleshooting

"VM not found" error

VM names are case-sensitive in vSphere. Use exact name from vmware-monitor inventory vms.

Guest exec returns empty output

Use run_vm_guest_command_with_output instead of run_vm_guest_command — it auto-captures stdout/stderr. Basic run_vm_guest_command only returns exit code.

Deploy OVA times out

Large OVA files (>10GB) may exceed the default 120s timeout. The upload happens via HTTP NFC lease — ensure network between the machine running vmware-aiops and ESXi is stable.

Plan apply fails mid-way

Run vmware-aiops plan list to see failed plan status. Ask user if they want to rollback with rollback_vm_plan. Irreversible steps (delete_vm) are skipped during rollback.

Connection refused / SSL error

1. Verify target is reachable: vmware-aiops doctor
2. For self-signed certs: set verify_ssl: false in config.yaml (lab environments only)

doctor reports "no vSphere role", or every command fails with "session is not authenticated"

The account logs in but has no vSphere role, so vCenter gives it no session (AIops#47). Assign it a role (Read-Only or higher) on the vCenter root folder: Administration > Access Control > Global Permissions. Changing the password will not help. vmware-aiops doctor reports this case by name, and attention counts such a target as unread (overall at least WARN).

Supported AI Platforms

Platform Comparison

MCP Server Integrations

The vmware-aiops MCP server works with any MCP-compatible agent or tool. Ready-to-use configuration templates are in examples/mcp-configs/.

Xiaoguai (小怪) — a self-hostable, audit-first agent platform (Rust, single binary + embedded SQLite) from the same maintainer. It runs the vmware-aiops MCP server as one of its toolboxes; being both an MCP consumer and an MCP server, its HMAC-chained audit log and human-on-the-loop approval gates line up with this skill's own audit + confirm design. See its MCP integration guide.

Fully local operation (no cloud API required):

# Aider + Ollama + vmware-aiops (via SKILL.md)aider --conventions skills/vmware-aiops/SKILL.md --model ollama/qwen2.5-coder:32b
# Any MCP agent + local model + vmware-aiops MCP server# See examples/mcp-configs/ for your agent's config format

Installation

Step 0: Prerequisites

# Python 3.10+ requiredpython3 --version
# Node.js 18+ required for Gemini CLI and Codex CLInode --version

Step 1: Clone & Install Python Backend

All platforms share the same Python backend.

git clone https://github.com/vmware-skills/VMware-AIops.gitcd VMware-AIopspython3 -m venv .venvsource .venv/bin/activatepip install -e .

Step 2: Configure

mkdir -p ~/.vmware-aiopscp config.example.yaml ~/.vmware-aiops/config.yaml# Edit config.yaml with your vCenter/ESXi targets

Set passwords via .env file (recommended):

# Use the templatecp .env.example ~/.vmware-aiops/.env
# Edit and fill in your passwords, then lock permissionschmod 600 ~/.vmware-aiops/.env

Security note: Prefer .env file over command-line export to avoid passwords appearing in shell history. The .env file should have chmod 600 (owner-only read/write).

Password environment variable naming convention:

VMWARE_{TARGET_NAME_UPPER}_PASSWORD# Replace hyphens with underscores, UPPERCASE# Example: target "home-esxi" → VMWARE_HOME_ESXI_PASSWORD# Example: target "prod-vcenter" → VMWARE_PROD_VCENTER_PASSWORD

Security Best Practices

• NEVER hardcode passwords in scripts or config files
• NEVER pass passwords as command-line arguments (visible in ps)
• ALWAYS use ~/.vmware-aiops/.env with chmod 600
• ALWAYS configure connections via config.yaml — credentials are loaded from .env automatically
• Config File Contents: config.yaml stores target hostnames, ports, and a reference to the .env file. It does not contain passwords or tokens. All secrets are stored exclusively in .env
• TLS: Enabled by default. Disable only for ESXi hosts with self-signed certificates in isolated lab environments
• Webhook: Disabled by default. When enabled, the daemon posts to your own configured URL only — no third-party service — every critical issue (alarms, events, ESXi log lines matching critical/panic/corrupt, targets it could not connect to) and every alarm/event warning — host-log warnings stay in the scan log, and info rows are never sent. Payloads carry the full issue text (entity names, alarm names, event messages, ESXi log excerpts, connection errors), which can include host names, IP addresses, and user names; they carry no credentials from your config or .env
• Least Privilege: Use a dedicated vCenter service account with minimal permissions. For monitoring-only use cases, prefer the read-only VMware-Monitor
• Prompt Injection Protection: All vSphere-sourced content is truncated, stripped of control characters, and wrapped in boundary markers before output
• Code Review: We recommend reviewing the source code and commit history before deploying in production
• Production Safety: For production environments, use the read-only VMware-Monitor instead. AI agents can misinterpret context and execute unintended destructive operations — real-world incidents have shown that AI-driven infrastructure tools without proper isolation can delete production databases and entire environments. VMware-Monitor eliminates this risk at the code level: no destructive functions exist in its codebase

Step 3: Connect Your AI Tool

Choose one (or more) of the following:

Option A: Claude Code

Method 1: Skills.sh or ClawHub (recommended)

Either installer places the skill in Claude Code's skills directory for you:

npx skills add vmware-skills/VMware-AIops# orclawhub install @zw008/vmware-aiops

Method 2: Manual skill install

git clone https://github.com/vmware-skills/VMware-AIops.gitcd VMware-AIops
# Copy the skill into Claude Code's personal skills directorymkdir -p ~/.claude/skills/vmware-aiopscp -r skills/vmware-aiops/. ~/.claude/skills/vmware-aiops/

For tool access (not just skill context), also register the MCP server:

claude mcp add vmware-aiops -- vmware-aiops mcp

Restart Claude Code, then:

> Show me all VMs on esxi-lab.example.com

Submit to Official Marketplace

This plugin can also be submitted to the Anthropic official plugin directory for public discovery.

Option B: Gemini CLI

# Install Gemini CLInpm install -g @google/gemini-cli
# Load the skill as project context (Gemini CLI reads GEMINI.md on startup)cp skills/vmware-aiops/SKILL.md ./GEMINI.md

For tool access (not just context), register the MCP server in ~/.gemini/settings.json:

{  "mcpServers": {    "vmware-aiops": {      "command": "vmware-aiops",      "args": ["mcp"],      "env": { "VMWARE_AIOPS_CONFIG": "~/.vmware-aiops/config.yaml" }    }  }}

Then start Gemini CLI:

gemini> Show me all VMs on my ESXi host

Option C: OpenAI Codex CLI

# Install Codex CLInpm i -g @openai/codex# Or on macOS:# brew install --cask codex
# Copy skill to Codex skills directorymkdir -p ~/.codex/skills/vmware-aiopscp skills/vmware-aiops/SKILL.md ~/.codex/skills/vmware-aiops/SKILL.md
# Copy AGENTS.md to project rootcp skills/vmware-aiops/SKILL.md ./AGENTS.md

Then start Codex CLI:

codex --enable skills> List all VMs on my ESXi

Option D: Aider (supports local models)

# Install Aiderpip install aider-chat
# Install Ollama for local models (optional)# macOS:brew install ollamaollama pull qwen2.5-coder:32b
# Run with cloud APIaider --conventions skills/vmware-aiops/SKILL.md
# Or with local model via Ollamaaider --conventions skills/vmware-aiops/SKILL.md \  --model ollama/qwen2.5-coder:32b

Option E: Continue CLI (supports local models)

# Install Continue CLInpm i -g @continuedev/cli
# Copy rules filemkdir -p .continue/rulescp skills/vmware-aiops/SKILL.md .continue/rules/vmware-aiops.md

Configure ~/.continue/config.yaml for local model:

models:  - name: local-coder    provider: ollama    model: qwen2.5-coder:32b

Then:

cn> Check ESXi health and alarms

Option F: Trae IDE

Copy the rules file to your project's .trae/rules/ directory:

mkdir -p .trae/rulescp skills/vmware-aiops/SKILL.md .trae/rules/project_rules.md

Trae IDE's Builder Mode reads .trae/rules/ Markdown files at startup.

Note: You can also install Claude Code extension in Trae IDE and use .claude/skills/ format directly.

Option G: Kimi Code CLI

# Copy skill file to Kimi skills directorymkdir -p ~/.kimi/skills/vmware-aiopscp skills/vmware-aiops/SKILL.md ~/.kimi/skills/vmware-aiops/SKILL.md

Option H: MCP Server (Glama / Claude Desktop)

The MCP server exposes VMware operations as tools via the Model Context Protocol. Works with any MCP-compatible client (Claude Desktop, Cursor, etc.).

After uv tool install vmware-aiops, start the MCP server with one command (v1.5.15+):

# Recommended — single command, no network re-resolvevmware-aiops mcp
# With a custom config pathVMWARE_AIOPS_CONFIG=/path/to/config.yaml vmware-aiops mcp

Claude Desktop config (claude_desktop_config.json):

{  "mcpServers": {    "vmware-aiops": {      "command": "vmware-aiops",      "args": ["mcp"],      "env": {        "VMWARE_AIOPS_CONFIG": "/path/to/config.yaml"      }    }  }}

# Run without installing (requires PyPI access each launch)uvx --from vmware-aiops vmware-aiops mcp
# Legacy entry point (still works, kept for backward compatibility)vmware-aiops-mcp

Option I: Standalone CLI (no AI)

# Already installed in Step 1source .venv/bin/activate
vmware-aiops vm power-on my-vm --target home-esxivmware-aiops deploy ova ./ubuntu.ova --name my-vm --target home-esxivmware-aiops datastore browse datastore1 --target home-esxi

Update / Upgrade

Already installed? Re-run the install command for your channel to get the latest version:

Check your current version: vmware-aiops --version

Chinese Cloud Models

For users in China who prefer domestic cloud APIs or have limited access to overseas services.

DeepSeek

Cost-effective, strong coding capability.

# Set DeepSeek API key (get from https://platform.deepseek.com)export DEEPSEEK_API_KEY="your-key"
# Run with Aideraider --conventions skills/vmware-aiops/SKILL.md \  --model deepseek/deepseek-coder

Persistent config ~/.aider.conf.yml:

model: deepseek/deepseek-coderconventions: skills/vmware-aiops/SKILL.md

Qwen (Alibaba Cloud)

Alibaba Cloud's coding model, free tier available.

# Set DashScope API key (get from https://dashscope.console.aliyun.com)export DASHSCOPE_API_KEY="your-key"
aider --conventions skills/vmware-aiops/SKILL.md \  --model qwen/qwen-coder-plus

Or via OpenAI-compatible endpoint:

export OPENAI_API_BASE="https://dashscope.aliyuncs.com/compatible-mode/v1"export OPENAI_API_KEY="your-dashscope-key"
aider --conventions skills/vmware-aiops/SKILL.md \  --model qwen-coder-plus-latest

Doubao (ByteDance)

export OPENAI_API_BASE="https://ark.cn-beijing.volces.com/api/v3"export OPENAI_API_KEY="your-ark-key"
aider --conventions skills/vmware-aiops/SKILL.md \  --model your-doubao-endpoint-id

With Continue CLI

Configure ~/.continue/config.yaml:

# DeepSeekmodels:  - name: deepseek-coder    provider: openai-compatible    apiBase: https://api.deepseek.com/v1    apiKey: your-deepseek-key    model: deepseek-coder
# Qwenmodels:  - name: qwen-coder    provider: openai-compatible    apiBase: https://dashscope.aliyuncs.com/compatible-mode/v1    apiKey: your-dashscope-key    model: qwen-coder-plus-latest

Local Models (Aider + Ollama)

For fully offline operation — no cloud API, no internet, full privacy.

Aider + Ollama + local Qwen/DeepSeek is ideal for air-gapped environments.

Step 1: Install Ollama

# macOSbrew install ollama
# Linux — download from https://ollama.com/download and install manually# See https://github.com/ollama/ollama for platform-specific instructions

Step 2: Pull a model

Hardware: 32B → ~20GB VRAM (or 32GB RAM for CPU). 7B → 8GB RAM.

Step 3: Run with Aider

pip install aider-chatollama serve
# Aider + local Qwen (recommended)aider --conventions skills/vmware-aiops/SKILL.md \  --model ollama/qwen2.5-coder:32b
# Aider + local DeepSeekaider --conventions skills/vmware-aiops/SKILL.md \  --model ollama/deepseek-coder-v2
# Low-memory optionaider --conventions skills/vmware-aiops/SKILL.md \  --model ollama/qwen2.5-coder:7b

Persistent config ~/.aider.conf.yml:

model: ollama/qwen2.5-coder:32bconventions: skills/vmware-aiops/SKILL.md

Local Architecture

User → Aider CLI → Ollama (localhost:11434) → Qwen / DeepSeek local model  │                                                    ↓  │                                          reads AGENTS.md instructions  │                                                    ↓  └──────────────────────────────→ vmware-aiops CLI ──→ ESXi / vCenter

Tip: Local models are fully offline — perfect for air-gapped environments or strict data compliance.

CLI Reference

# Diagnosticsvmware-aiops doctor                   # Check environment, config, connectivityvmware-aiops doctor --skip-auth       # Skip vSphere auth check (faster)vmware-aiops access-info [--tool <mcp-tool>] [--target <name>]  # Account in use + privileges per tool
# MCP Config Generatorvmware-aiops mcp-config generate --agent goose        # Generate config for Goosevmware-aiops mcp-config generate --agent claude-code  # Generate config for Claude Codevmware-aiops mcp-config list                          # List all supported agents
# VM operationsvmware-aiops vm power-on my-vm                                 # Power onvmware-aiops vm power-off my-vm                                # Graceful shutdown (2x confirm)vmware-aiops vm power-off my-vm --force                        # Force power off (2x confirm)vmware-aiops vm create my-new-vm --cpu 4 --memory 8192 --disk 100  # Create VMvmware-aiops vm delete my-vm                                   # Delete VM (asks twice; --dry-run previews)vmware-aiops vm reconfigure my-vm --cpu 4 --memory 8192        # Reconfigure (2x confirm)vmware-aiops vm snapshot-create my-vm --name "before-upgrade"  # Create snapshotvmware-aiops vm snapshot-list my-vm                            # List snapshotsvmware-aiops vm snapshot-revert my-vm --name "before-upgrade"  # Revert snapshotvmware-aiops vm snapshot-delete my-vm --name "before-upgrade"  # Delete snapshot (waits ≤30 min for consolidation)vmware-aiops vm snapshot-delete my-vm --name "old-big" --no-wait  # Fire async, return a task idvmware-aiops vm task-status task-1234                          # Poll an async task by idvmware-aiops vm clone my-vm --new-name my-vm-clone             # Clone VMvmware-aiops vm migrate my-vm --to-host esxi-02                # vMotionvmware-aiops vm set-ttl my-vm --minutes 60                     # Auto-delete in 60 minvmware-aiops vm cancel-ttl my-vm                               # Cancel TTLvmware-aiops vm list-ttl                                       # Show all TTLsvmware-aiops vm clean-slate my-vm --snapshot baseline          # Revert to baseline (2x confirm)
# Guest Operations (requires VMware Tools in guest)vmware-aiops vm guest-exec my-vm --cmd /bin/bash --args "-c 'whoami'" --user rootvmware-aiops vm guest-upload my-vm --local ./script.sh --guest /tmp/script.sh --user rootvmware-aiops vm guest-download my-vm --guest /var/log/syslog --local ./syslog.txt --user root
# Plan → Apply (multi-step operations)vmware-aiops plan list                                        # List pending/failed plans
# Deployvmware-aiops deploy ova ./ubuntu.ova --name my-vm --datastore ds1      # Deploy from OVAvmware-aiops deploy template golden-ubuntu --name new-vm               # Deploy from templatevmware-aiops deploy linked-clone --source base-vm --snapshot clean --name test-vm  # Linked clone (seconds)vmware-aiops deploy iso my-vm --iso "[datastore1] iso/ubuntu-22.04.iso"  # Attach ISOvmware-aiops deploy mark-template golden-vm                            # Convert VM to templatevmware-aiops deploy batch-clone --source base-vm --count 5 --prefix lab  # Batch clonevmware-aiops deploy batch deploy.yaml                                  # Batch deploy from YAML spec
# Clustervmware-aiops cluster info my-cluster                                   # Cluster details (HA/DRS status)vmware-aiops cluster create my-cluster --ha --drs                      # Create cluster with HA+DRSvmware-aiops cluster delete my-cluster                                 # Delete cluster (2x confirm)vmware-aiops cluster add-host my-cluster --host esxi-03                # Add host to cluster (2x confirm)vmware-aiops cluster remove-host my-cluster --host esxi-03             # Remove host (2x confirm)vmware-aiops cluster configure my-cluster --ha --drs                   # Configure HA/DRS (2x confirm)
# Alarm managementvmware-aiops alarm list                                                # List triggered alarmsvmware-aiops alarm acknowledge esxi-01 "Host memory usage"             # Acknowledge alarmvmware-aiops alarm reset esxi-01 "Host memory usage"                   # Clear alarms (2x confirm; clears ALL matching entity type + status)
# Datastore (browse and scan only — iSCSI/vSAN moved to vmware-storage)vmware-aiops datastore browse datastore1 --path "iso/"                 # Browse datastorevmware-aiops datastore scan-images --target home-esxi                  # Scan all datastores for images
# Scanvmware-aiops scan now              # One-time scan of alarms and events (host logs: daemon only)
# Daemonvmware-aiops daemon start          # Start scannervmware-aiops daemon status         # Check statusvmware-aiops daemon stop           # Stop daemon
# Companion skills for other operations:#   vmware-monitor: inventory, alarms, events, sensors#   vmware-storage: datastores, iSCSI, vSAN#   vmware-vks:     Tanzu/TKC cluster lifecycle

Configuration

See config.example.yaml for all options.

Project Structure

VMware-AIops/├── skills/                        # Skills index (npx skills add)│   └── vmware-aiops/│       ├── SKILL.md               # Slimmed-down skill (progressive disclosure)│       └── references/            # Detailed docs loaded on-demand│           ├── capabilities.md    # Full capabilities tables│           ├── cli-reference.md   # Complete CLI reference│           └── setup-guide.md     # Install, security, AI platforms├── vmware_aiops/                  # Python backend│   ├── config.py                  # YAML + .env config│   ├── connection.py              # Multi-target pyVmomi│   ├── cli/                       # Typer CLI (double confirm)│   ├── ops/                       # Operations│   │   ├── inventory.py           # VMs, hosts, datastores, clusters│   │   ├── health.py              # Alarms, events, sensors│   │   ├── vm_lifecycle.py        # VM CRUD, snapshots, clone, migrate│   │   ├── vm_deploy.py           # OVA, template, linked clone, batch deploy│   │   └── datastore_browser.py   # Datastore browsing, image discovery│   ├── scanner/                   # Log scanning daemon│   ├── notify/                    # Notifications (JSONL + webhook)│   └── mcp_server/                # MCP server wrapper│       ├── server.py              # FastMCP server with tools│       └── __main__.py├── examples/mcp-configs/          # MCP client config templates├── tests/                         # Test suite├── smithery.yaml                  # Smithery marketplace config├── RELEASE_NOTES.md├── config.example.yaml└── pyproject.toml

API Coverage

Built on pyVmomi (vSphere Web Services API / SOAP).

Related Projects

Troubleshooting & Contributing

If you encounter any errors or issues, please send the error message, logs, or screenshots to [email protected]. Contributions are welcome — feel free to join us in maintaining and improving this project!

Remote access (HTTPS) and limiting tools

By default vmware-aiops mcp speaks stdio to one local client and offers every tool. Two options change that (vmware-policy ≥ 1.21.0).

Serve over HTTPS so clients on other machines can connect. Every caller needs a bearer token:

vmware-audit token create alice          # prints the token once; give it to alicevmware-audit token list                  # names onlyvmware-audit token revoke alice          # refused from its next request, no restartvmware-aiops mcp --transport https --host 0.0.0.0 --port 8443 \    --cert /etc/vmware-mcp/server.pem --key /etc/vmware-mcp/server-key.pem \    --public-name mcp.example.com

Connect a client (Claude Code shown; any MCP client that supports streamable HTTP and custom headers works):

claude mcp add --transport http vmware-aiops https://mcp.example.com:8443/mcp \    --header "Authorization: Bearer $VMWARE_MCP_TOKEN"

{  "mcpServers": {    "vmware-aiops": {      "type": "http",      "url": "https://mcp.example.com:8443/mcp",      "headers": { "Authorization": "Bearer ${VMWARE_MCP_TOKEN}" }    }  }}

• The server will not start without a token, refuses Host/Origin values outside --public-name, and checks before it binds that unauthenticated requests get 401. Use a certificate your clients trust for those names.
• Behind a TLS-terminating reverse proxy on the same machine use --transport http --host 127.0.0.1 --public-name mcp.example.com and pass the Authorization header through.
• Behind a load balancer or gateway on another machine, keep --transport https (a self-signed certificate is fine), list in --public-name the host name the balancer sends as Host, and have it forward Authorization. Its health probe can use GET /healthz, which needs no token and returns only {"status": "ok"} (vmware-policy ≥ 1.23.0).
• Send traceparent or X-Request-ID with each request to have it stored on the audit row; vmware-audit log --correlation-id ID then finds the calls behind one platform request.
• Tools that read or write files on this server (here: guest file upload/download, OVA and deployment-spec files) never touch its configuration, tokens, audit log or TLS key, and over the network they are off unless VMWARE_MCP_FILE_ROOT names a directory for them (vmware-policy ≥ 1.23.1).
• The audit log records each caller as token:<name>. Calls run side by side, with up to VMWARE_MCP_POOL_SIZE (default 2) sessions per target; each session is a login on the target, and a call waits at most 300 s for one.
• Everyone who can reach the endpoint acts with this server's account. Tokens identify callers; they grant nothing on the platform. Run one endpoint per privilege level.
• Several people or roles on one endpoint: run vmware-audit roles enable once, then give each token a role (vmware-audit token create NAME --role reader, or a role in roles.yaml naming tools and targets; vmware-policy ≥ 1.24.0). A role limits what the token may call on this endpoint; it cannot grant more than the account has. See vmware-policy's README, "Roles for network callers".

Offer only read tools, plus the writes you choose:

vmware-aiops mcp --expose read --allow-write acknowledge_vcenter_alarm# or: VMWARE_AIOPS_MCP_EXPOSE=read VMWARE_AIOPS_MCP_ALLOW_WRITE=acknowledge_vcenter_alarm vmware-aiops mcp

Tools not offered do not appear in the client's tool list. A bad value or an unknown write tool stops the server (exit 2). This limits what clients are offered; it is not authorization, so the account itself should be least-privilege.

Details: skills/vmware-aiops/references/setup-guide.md ("Serving MCP over HTTPS", "Limiting which tools are offered").

License

MIT