# ComputeInstances Object

> Source: https://truto.one/docs/api-reference/unified-cloud-infrastructure-api/computeinstances/

Schema for the `ComputeInstances` resource in **Unified Cloud Infrastructure API**.

## Properties

- **`id`** _(string, required)_
  Truto's stable unified identifier for this object. Opaque; use provider_id to address the object in the provider's own console or API.
- **`provider_id`** _(string)_
  The provider's own identifier, passed through verbatim — a full ARN, resource id, or self-link. Never truncated, prefixed, or normalized.
- **`provider`** _(string)_
  Which cloud this object was read from.
  Allowed: `aws`, `azure`, `gcp`
- **`account`** _(string)_
  The account, subscription or project this object belongs to.
- **`region`** _(string)_
  Where the object is located. 'global' denotes genuinely global resources (e.g. IAM, GCP VPC networks) rather than a guessed region. A substituted region (when none is derivable) is flagged in unreadable_fields.
- **`native_type`** _(string)_
  The provider's own type string, unmodified — for example 'AWS::S3::Bucket', 'aws_iam_role', 'Microsoft.Sql/servers'. Used for display and drill-down.
- **`tags`** _(object)_
  Key-value pairs exactly as the customer set them: no case folding, key/value normalization, or merging of separate provider concepts. An empty object means no tags; tags that could not be read appear in unreadable_fields instead.
- **`collected_at`** _(string)_
  When Truto actually read this object from the provider, in UTC ISO 8601 with the offset present — not the request time, and not a provider timestamp.
- **`updated_at`** _(string)_
  When the object was last modified at the provider, in UTC ISO 8601. Null when the provider does not record one - see unreadable_fields.
- **`created_at`** _(string)_
  When the object was created at the provider, in UTC ISO 8601. Null when the provider does not record one - see unreadable_fields.
- **`unreadable_fields`** _(array<object>)_
  Fields on this object that could not be read, and why. An empty array means everything was read. Use this to tell a real value from a missing one.
  - **`field`** _(string)_
    The property on this resource that could not be read.
  - **`reason`** _(string)_
    Why the field could not be read. available_on_get: Not read on list to keep the list fast; call GET for this resource's id to get it.
    Allowed: `not_supported_by_provider`, `not_configured`, `permission_denied`, `not_collected`, `collection_error`, `partially_collected`, `available_on_get`
  - **`detail`** _(string)_
    Explanation of the reason, where one adds anything.
- **`remote_data`** _(object)_
  Raw data returned from the remote API call.
- **`instance_id`** _(string)_
  The provider's own identifier for the machine, verbatim (an EC2 i- id, an Azure VM resource id, a GCP instance id).
- **`name`** _(string)_
  The display name: on AWS the Name tag, which is optional, so null means the instance has no Name tag.
- **`state`** _(string)_
  Lifecycle state, normalised. On AWS shutting-down is 'terminating'. Terminated EC2 instances stay listed for about an hour. The provider's own value is in remote_data.
  Allowed: `pending`, `running`, `stopping`, `stopped`, `terminating`, `terminated`, `unknown`
- **`instance_type`** _(string)_
  The provider's size or machine type, verbatim (t3.micro, Standard_D2s_v3, e2-medium).
- **`os_family`** _(string)_
  Operating system family. linux_unix covers Linux and other Unix-like systems; unknown when the provider does not say. The provider's own value is in remote_data.
  Allowed: `windows`, `linux_unix`, `macos`, `unknown`
- **`private_ip_addresses`** _(array<string>)_
  Every private IPv4 address on any of the machine's network interfaces.
- **`public_ip_addresses`** _(array<string>)_
  Every public IPv4 address on any network interface, including addresses on secondary private IPs. Empty means no public IPv4.
- **`ipv6_addresses`** _(array<string>)_
  IPv6 addresses on any network interface. These are globally unique; whether they are reachable from the internet depends on routing and firewall rules.
- **`has_public_ip`** _(boolean)_
  Whether any network interface has a public IPv4 address.
- **`network_boundary`** _(object)_
  The VPC / virtual network the machine sits in.
  - **`id`** _(string)_
    The target's `id`.
- **`subnet_id`** _(string)_
  The subnet of the primary network interface, verbatim.
- **`availability_zone`** _(string)_
  The zone the machine runs in.
- **`attached_rule_sets`** _(array<object>)_
  Security groups on the machine and on every one of its network interfaces.
  - **`id`** _(string)_
    The target's `id`.
- **`machine_identity_id`** _(string)_
  The identity the machine runs as, verbatim: on AWS the instance profile ARN (an instance profile holds one IAM role; it is not the role ARN workload_identities is keyed on), on Azure the managed identity, on GCP the service account. Null when the machine has none.
- **`metadata_service`** _(object)_
  The instance metadata service (IMDS) settings. Null when the provider did not return them - see unreadable_fields.
  - **`enabled`** _(boolean)_
    Whether the metadata endpoint answers at all.
  - **`token_required`** _(boolean)_
    Whether every metadata request must carry a session token. On AWS true is IMDSv2-only (httpTokens required); false still answers IMDSv1 requests without a token.
  - **`hop_limit`** _(integer)_
    How many network hops a metadata response may travel. Above 1, containers on the host can reach the endpoint.
- **`image_id`** _(string)_
  The image the machine was launched from, verbatim (an AMI id on AWS).
- **`image`** _(object)_
  The image the machine was launched from. machine_images lists only images the account owns, so an Amazon, Marketplace or other shared image does not resolve there; GET machine_images/{id} still reads it.
  - **`id`** _(string)_
    The target's `id`.
- **`launched_at`** _(string)_
  When the machine last started. On AWS this moves on every stop and start, so it is not a creation date.
- **`is_managed`** _(boolean)_
  Whether a fleet-management service (AWS Systems Manager, Azure Arc or Update Manager, GCP VM Manager) currently registers the machine as a managed node. On AWS, Systems Manager does not report stopped instances, so a stopped instance reads false even when it is registered. Read on GET only.
- **`managed_instance`** _(object)_
  The managed-node record for this machine, when is_managed is true. Read on GET only.
  - **`id`** _(string)_
    The target's `id`.

## Methods

- [GET /unified/cloud-infrastructure/compute_instances/{id}](/docs/api-reference/unified-cloud-infrastructure-api/computeinstances/get) — Get Compute instances
- [GET /unified/cloud-infrastructure/compute_instances](/docs/api-reference/unified-cloud-infrastructure-api/computeinstances/list) — List Compute instances