Hanzo AI
OpenapiCompute

List nodes

Returns every DOKS worker node in the org's clusters as a machine — the SAME set the fleet folds in (managedMachines), exposed directly under the k8s…

GET /v1/compute/k8s/nodes

Addresshttps://api.hanzo.ai/v1/compute/k8s/nodes
MethodGET
OperationlistKubernetesNodes
AuthAuthorization: Bearer $HANZO_API_KEY

Returns every DOKS worker node in the org's clusters as a machine — the SAME set the fleet folds in (managedMachines), exposed directly under the k8s noun. House account (hanzo-org cluster tag) + BYOC, deduped by Visor.

Request

GET /v1/compute/k8s/nodes takes no parameters and no body — the credential is the whole request.

Response

StatusBodyMeaning
200compute.nodeListok
defaultproblem-detailsrefused

200 body — 29 fields.

FieldInTypeAlwaysDescription
nodesbodycompute.machineView[]Nodes is one row per worker node, in the SAME machineView shape the machines surface emits — a node IS a machine.
nodes[].agentbodystringAgent is the cloud Agent this machine runs, lifted out of the binding so a list reads without following one.
nodes[].bindingbodycompute.agentBinding
nodes[].binding.agentNamebodystringAgentName is the cloud Agent (/v1/agent) this machine runs — the agent a message to the bot is actually run against.
nodes[].binding.botVersionbodystringBotVersion pins the @hanzo/bot runtime version the machine runs.
nodes[].binding.createdTimebodystringCreatedTime is when the binding was first made.
nodes[].binding.machineIdbodystringMachineId is the bound machine as vm addresses it, owner-qualified ("<org>/<machine>").
nodes[].binding.messagebodystringMessage is vm's human-readable detail on Status ("machine provisioning; @hanzo/bot runtime not yet confirmed") — the reason behind the state, not a second…
nodes[].binding.namebodystringName is the binding's own key, which is the machine's id: a machine hosts at most one agent, so the binding is named for it.
nodes[].binding.orgbodystringOrg is the Hanzo tenant the binding belongs to.
nodes[].binding.ownerbodystringOwner is the tenant vm filed the binding under, resolved from the ?owner it was called with — which is the caller's validated org and never a body field.
nodes[].binding.providerbodystringProvider is the cloud the bound machine runs on, carried here so a bindings list says where each bot lives without a second read per machine.
nodes[].binding.publicIpbodystringPublicIp is the bound machine's public address as vm recorded it on the binding.
nodes[].binding.statusbodystringStatus is the binding's lifecycle in VM's OWN words — "Pending" while the machine provisions and the runtime is unconfirmed, "running" once vm has confirmed…
nodes[].binding.updatedTimebodystringUpdatedTime is when vm last reconciled it — the age of Status.
nodes[].createdTimebodystringCreatedTime is when the machine came into being: the provider's own creation timestamp for a Visor machine, passed through in whatever form it states it, and…
nodes[].gpubodystringGPU names the accelerators this machine holds ("H100", or "2× NVIDIA GB10" for a BYO machine reporting a matched pair).
nodes[].idbodystringID addresses this machine on the /v1/compute/machines/:id routes: the org-scoped NAME Visor keys a machine by, falling back to the provider id for a machine…
nodes[].imagebodystringImage is the OS image the machine booted from, as the provider names it.
nodes[].membodystringMem is system RAM rendered for a human ("8 GB"), not a number to compute with.
nodes[].namebodystringName is the label to show a human — Visor's displayName, or the machine name when it carries none. A BYO machine's is its hostname.
nodes[].osbodystringOs is the operating system on the machine — Visor's record for a provisioned one, the host's own report (linux, darwin, windows) for a BYO one.
nodes[].privateIpbodystringPrivateIp is the address on the provider's own network, reachable from the org's other machines in the same region.
nodes[].providerbodystringProvider is the cloud that runs the machine ("digitalocean"), or "byo" for one the operator dialed in with hanzo link.
nodes[].publicIpbodystringPublicIp is the internet-facing address the provider assigned.
nodes[].regionbodystringRegion is the provider region slug ("sfo3"), or the zone when the provider reports only that.
nodes[].statusbodystringStatus is the lifecycle state in the PROVIDER's own words ("active", "running", "off"), passed through rather than mapped onto a vocabulary of ours.
nodes[].typebodystringType is the provider SIZE SLUG the machine runs at ("s-2vcpu-4gb", "gpu-h100x8-640gb") — the value a launch asks for, and what Vcpu/Mem/GPU are read out of…
nodes[].vcpubodyinteger (int64)Vcpu is logical cores — the provider's own cpuSize when that is a clean integer, else the count read out of the size slug (4 from "s-4vcpu-8gb").

Failure carries the platform error shape — see Errors.

Examples

hanzo compute k8s nodes

Compute API · All Hanzo APIs · Interactive reference

How is this guide?

On this page