Skip to main content
GET
Error

Response

200 - application/json

The service is serving.

Deliberately terse. The backend inventory is reconnaissance material and is never reported on an unauthenticated endpoint.

ok
boolean
required

True whenever the process is serving.

service
string
required

A constant service identifier.

status
enum<string>
required

ONE WORD about the whole service, so a reader does not have to invent its own definition of down from the per subsystem booleans below. degraded means at least one of those booleans is false; incident means an operator declared one, and a declaration wins over a clean set of loops because a person can see faults this process cannot. An absent subsystem is a configuration and never counts against this word.

Available options:
ok,
degraded,
incident
incident
object

Present only while an operator has declared an incident. A half declaration (a sentence with no link) is discarded rather than published, because a banner whose link goes nowhere teaches people to ignore banners.

statusPage
string

The public status page, when this deployment has one. Published so a reader links to it without hardcoding an address, and absent rather than guessed when there is none.

revision
string

The exact build this process is running, as an opaque identifier. Published because it is the only way to tell a deployment that finished from one that is still rolling, and absent when nothing was baked into the image rather than guessed at.

billing
object

Present only on a hosted deployment that runs the billing loop. viewC appears only while a billing migration is outstanding and names which way it is outstanding.

frames
object

Present only where the frame retention sweep runs.

webhooks
object

Present only where the webhook delivery loop runs.

fulfilmentUnwind
object

Present only where the loop that unwinds a paid order that never arrived runs.

fleetIntegrity
object

Whether every device this deployment sells is one the provider still has, and none of them reaches the end of its paid provider term within the alert window (72 hours by default) without being renewed. ABSENT when the verdict is unknown, which is why a reader must treat a missing key as 'not answered' and never as 'fine': staging omits it today and production does not.

email
object

Whether a notification can reach a person by mail. False is a real and ordinary state: the bell still fills, and nobody is told while the console is shut. Not counted into status.

cloudReadyEmail
object

Whether the worker that emails the buyer when a Cloud phone is ready is running in this process. False is the ordinary state until delivery is switched on. Not counted into status.

liveVideo
object

Whether live video for robot-driven phones is switched on here. Never counted into status.

latencyProbes
object

Whether the rig latency probe results are available here. Present only on a deployment with a database. Never counted into status.

opsAlerts
object

Whether operations alerts reach a person by email here. Never counted into status.

ownerMoney
object

Hosted only. A configuration, never counted into status.

staffDirectory
object

Hosted only. A configuration, never counted into status.

staffOrgs
object

Hosted only. A configuration, never counted into status. A state only, never an organisation id.

schedules
object

The customer-facing schedule sweeper. disabled is a configuration and never degrades status.

provisioning
object

The loop that puts a cloud phone in a customer's hands. disabled means every such order waits for a person, which is why absence is PUBLISHED here rather than left as a missing key.

deviceTakes
object

The loop that takes a device back when a session ends without releasing it. disabled means a session that ends badly keeps its machine.

cloudBillingScan
object

The hosted missed-Stripe-webhook Cloud billing scan. disabled means no periodic recovery pass runs on this instance.

cloudRenewalScan
object

The hosted managed Cloud term scan. disabled means no periodic pass runs; running can be observation-only, with supplier spending and period-end billing stop separately switched off.

rigTelemetry
object

Present only on a hosted deployment with a database, where robot arm health samples are kept. Not counted into status.

opsStream
enum<string>

How this replica hears the operator console's live events. listen: a checked database notification channel delivers them as they happen. polling: they are read every 5 seconds, which is also what serves when no notification channel is configured. Present only where the operator surface runs with a database. Never counted into status.

Available options:
listen,
polling
tenancy
object

Deployment posture, on a hosted deployment only. Enums throughout and never a timestamp, an issuer name or a count: a reader learns which gates are armed, not when one opens.