VMware Storage

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 — zhouwei008@gmail.com 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

VMware vSphere storage management: datastores, iSCSI, vSAN, and read-only Fibre Channel / multipath diagnostics — 16 MCP tools, domain-focused and lightweight.

Split from vmware-aiops for lighter context and local model compatibility.

MCP uses local stdio by default. Opt-in HTTPS opens the configured listener and requires TLS plus caller authentication; it is a network service, not a local-only mode. Credential examples create owner-only placeholder files without overwriting existing files; edit secrets locally.

Container examples build a local version-tagged image and run it with --pull=never, avoiding registry fallback. A local version tag is not an immutable image digest.

Upgrading to 2.0

2.0 renames 10 of its MCP tools (all but the four that already started with a verb) so every name starts with a verb (e.g. fc_adapter_list → list_fc_adapters, storage_device_paths → list_storage_device_paths, storage_iscsi_add_target → add_iscsi_target). The full old → new table is in RELEASE_NOTES.md. Three tools also change a parameter: browse_datastore and scan_datastore_images take datastore_name (was ds_name, now the same as vmware-aiops' browse_datastore), and list_all_datastores gains an optional limit (same as vmware-monitor's). Everything else stays: same tools, safety gates and CLI commands (the CLI still takes the datastore name as before). If an agent prompt, script or ~/.vmware/rules.yaml names a tool, update it — 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.

Companion Skills

Quick Install

# Via PyPI
uv tool install vmware-storage==2.6.1

# Or pip
pip install vmware-storage==2.6.1

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-storage

Configuration

mkdir -p ~/.vmware-storage
cp config.example.yaml ~/.vmware-storage/config.yaml
# Edit with your vCenter/ESXi credentials

(
  umask 077
  set -C  # Refuse existing files; edit an existing .env locally instead
  cat > ~/.vmware-storage/.env <<'EOF'
VMWARE_MY_VCENTER_PASSWORD=REPLACE_IN_LOCAL_EDITOR
EOF
)
# Before continuing, edit .env locally; never type live secrets into shell commands.

# Verify
vmware-storage doctor

This writes placeholders only. Enter passwords or tokens using a local editor after creation; do not paste live credentials into terminal commands, chat or commits. If the file exists, the command refuses to overwrite it; edit the existing file after checking its permissions are 600.

MCP Tools (16)

Auto-Remediation Patterns (PoC)

The patterns/ directory hosts L5 auto-remediation candidate patterns from the Enterprise Harness Engineering framework. The first PoC pattern, patterns/iscsi-target-stale-rescan.yaml, describes an iSCSI HBA rescan as a low-risk, reversible, repeatable operation. The pattern schema is documented here only — runtime enforcement is not yet wired up, so this is a reference design, not production auto-remediation.

Common Workflows

Set Up iSCSI Storage on a Host

1. Enable iSCSI adapter: vmware-storage iscsi enable esxi-01
2. Add target: vmware-storage iscsi add-target esxi-01 10.0.0.100
3. Verify: vmware-storage iscsi status esxi-01

The add-target command automatically rescans storage. Use --dry-run to preview any write command first.

Find Deployable Images Across Datastores

For scripts, add --json to datastore list, datastore browse, or datastore scan-images. Successful output is the complete result envelope, including items, counts, truncation and any cache note. Without the flag, list keeps its table and browse/scan-images keep their legacy JSON arrays. Other read commands already output JSON by default. Failures still exit nonzero with the existing CLI error message.

1. List all datastores: vmware-storage datastore list
2. Scan for images: vmware-storage datastore scan-images datastore01
3. Browse with a pattern: vmware-storage datastore browse datastore01 --pattern "*.iso"

vSAN Health Assessment

1. Check health: vmware-storage vsan health Cluster-Prod
2. Check capacity: vmware-storage vsan capacity Cluster-Prod
3. If issues found, investigate with vmware-monitor for alarms and events

Check Fibre Channel Paths

1. Dead or disabled paths behind a datastore: vmware-storage paths devices --datastore ds-fc-01
2. Hosts that see a shared device through a different number of paths than their peers, or not at all: vmware-storage paths devices --cluster Cluster-Prod --only-differences
3. WWPNs for the SAN team: vmware-storage paths fc-adapters --cluster Cluster-Prod

Hosts that could not be read are listed in hosts_not_read and are never reported as missing a device. Path states are reported as vSphere reports them — standby is not flagged.

CLI

# Datastore
vmware-storage datastore list
vmware-storage datastore browse datastore01
vmware-storage datastore scan-images datastore01

# iSCSI
vmware-storage iscsi status esxi-01
vmware-storage iscsi enable esxi-01
vmware-storage iscsi add-target esxi-01 192.168.1.100
vmware-storage iscsi remove-target esxi-01 192.168.1.100
vmware-storage iscsi rescan esxi-01

# vSAN
vmware-storage vsan health Cluster-Prod
vmware-storage vsan capacity Cluster-Prod

# Fibre Channel / multipath (read-only)
vmware-storage paths fc-adapters --cluster Cluster-Prod
vmware-storage paths devices --datastore ds-fc-01

# Diagnostics
vmware-storage doctor

MCP Server

After uv tool install vmware-storage==2.6.1, start the MCP server with one command (v1.5.15+):

# Recommended — single command, no network re-resolve
vmware-storage mcp

# With a custom config path
VMWARE_STORAGE_CONFIG=/path/to/config.yaml vmware-storage mcp

# Or via Docker
docker compose up -d

Agent Configuration

Add to your AI agent's MCP config:

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

# Run without installing (requires PyPI access each launch)
uvx --from vmware-storage==2.6.1 vmware-storage mcp

# Legacy entry point (still works, kept for backward compatibility)
vmware-storage-mcp

Why a Separate Skill?

vmware-aiops has 61 MCP tools — too heavy for local LLMs (7B-14B). By splitting storage into its own skill:

• 16 tools — fits comfortably in small model context windows
• Domain-focused — storage admins get only what they need
• Least privilege — can run under a vCenter service account with storage read-only permissions
• Composable — use alongside vmware-monitor or vmware-aiops as needed

Version Compatibility

Python: 3.10+ (since v1.5.27 — previously 3.11+). Tested on 3.10 / 3.11 / 3.12.

Official Broadcom References

• SDKs: https://developer.broadcom.com/sdks — VCF Python SDK, vSAN Management SDK (bundled in pyVmomi)
• REST APIs: https://developer.broadcom.com/xapis — vSAN Management API, VCF API
• CLI Tools: https://developer.broadcom.com/tools — PowerCLI 9.1, ESXCLI

Safety

Troubleshooting

Remote access (HTTPS) and limiting tools

By default vmware-storage 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 alice
vmware-audit token list                  # names only
vmware-audit token revoke alice          # refused from its next request, no restart
vmware-storage 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-storage https://mcp.example.com:8443/mcp \
    --header "Authorization: Bearer $VMWARE_MCP_TOKEN"

{
  "mcpServers": {
    "vmware-storage": {
      "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.
• The audit log records each caller as token:<name>. Calls run side by side, also on the same target: the server keeps up to VMWARE_MCP_POOL_SIZE (default 2) sessions per target, each one a login on that target, and a call waits at most 300 s for a free session (vmware-policy ≥ 1.22.0).
• 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. vmware-storage access-info shows which account that is and what it may do.
• 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-storage mcp --expose read --allow-write enable_iscsi
# or: VMWARE_STORAGE_MCP_EXPOSE=read VMWARE_STORAGE_MCP_ALLOW_WRITE=enable_iscsi vmware-storage 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-storage/references/setup-guide.md ("Serving MCP over HTTPS", "Limiting which tools are offered").

License

MIT