VPC networks
The VPC networks list shows the project’s VPC networks, with their subnet mode, routing mode and subnet count.
For VPC networks, r reloads the list. The equivalent gcloud command is:
gcloud --configuration <active-config> --project <project> \ compute networks list --format=json --limit=500This is one flat, project-wide call. A network is a global resource, so there is no region to iterate and no second call to make.
The columns are NAME, MODE, ROUTING and SUBNETS. MODE is AUTO, CUSTOM or
LEGACY. ROUTING is REGIONAL or GLOBAL. SUBNETS is how many subnets the
network has.
nuvem derives MODE and ROUTING from the API’s own fields,
autoCreateSubnetworks and routingConfig.routingMode. It never reads the
x_gcloud_subnet_mode and x_gcloud_bgp_routing_mode keys that gcloud adds to
the same JSON. Those two keys carry exactly the strings this table shows, but
they are gcloud’s presentation layer and not API fields. The x_ prefix marks
them as gcloud’s own, and gcloud makes no promise to keep them.
LEGACY is a network that omits autoCreateSubnetworks altogether. It stays
distinct from CUSTOM because the two differ in what you do next: a custom
network has subnets to go and look at, and a legacy network cannot have any. A
network whose JSON carries no routingConfig leaves ROUTING blank. It does not
read REGIONAL. Regional is only the API’s default for a network created
without one, so nuvem does not assume it for a listing that did not say.
SUBNETS is a count, not a list. The subnetworks field is an array of full
self-link URLs, 43 of them on the default network of a fresh project, which
does not fit in a table cell. The subnets have a table of their own.
A count of 0 shows as 0 and never as a dash, because the API listed the
network’s subnets and found none. A custom-mode network with no subnets is
worth noticing, so the cell states it.
A network’s description has no column, because it is unbounded free text that would push the four columns off a narrow terminal. It is searchable, and the preview pane shows it in full.
/ narrows wholly client-side here and sends nothing to the server. Three of
the four columns are derived values that appear verbatim in no raw field, so
a filter sent to the server would drop rows you can read on screen. Label
filtering (-l k=v) reports unsupported, because the compute/v1 networks
resource has no labels field at all.
Because the narrowing is client-side, applying a filter makes no call at all. It narrows the rows already in memory again, which also protects the table. A list refresh that fails clears the table and its context. A table refreshed on every filter keystroke would go blank on a short auth or quota error while you type. r still fetches, because r is an explicit request for fresh rows.
The networks table is list-only and read-only. Enter does nothing on a row, there is no detail pane, and nuvem never creates, deletes or modifies a network. L is refused, not scoped to the selected network. Cloud Logging attributes VPC flow logs and firewall hit logs to the subnetwork, so a network-scoped query would return zero rows when you expect many.
There is no gateway column: gatewayAddress is a subnet field, and a
network carries no address of its own.
The fixture behind demo mode and the tests, tests/,
is a live capture taken on 2026-08-25 with the project id replaced by
nuvem-fixture-project. Its field set and key spellings are the API’s own.
The CUSTOM, GLOBAL and LEGACY cases are not in it, because the account it
was captured from can reach no such network. Those arms are pinned against
rows constructed from Google’s published compute/v1/networks reference, and
the tests that use them say so.
nuvem is released under the MIT License. The source is on GitHub.
nuvem is an independent project, not affiliated with or endorsed by Google. Google Cloud is a trademark of Google LLC.