Ornn CLI

Install @ornncompute/cli to browse capacity, launch reserved compute, connect, and inspect health from a terminal. The installed command is ornn. To integrate without the CLI, call the REST API with an API key. CLI and REST equivalents lists the request behind each command.

Install

Use Node.js 20 or newer:

npm install -g @ornncompute/cli
ornn --version

Or use the hosted installer:

curl -fsSL https://compute.ornn.com/cli/install | sh

The installer adds the npm global bin directory to your shell profile when needed. Restart the shell, then sign in.

ornn login
ornn context

ornn login opens the Ornn browser device flow. On a remote terminal, print the URL without opening a browser:

ornn login --no-browser

Command	What it does
`ornn login [--organization <exact-name>] [--no-browser] [--timeout <seconds>]`	Approve a sign-in in the browser and return to the terminal.
`ornn context [--organization <exact-name-or-id>]`	Show the user, selected organization, approval, and available organizations, or select another organization.
`ornn logout`	End the remote session and remove the local credential.

The CLI stores the approved session in auth.json under ~/.config/ornn, readable only by you. Set ORNN_CONFIG_HOME to keep it somewhere else, for example a separate directory per CI job. A session ends when you run ornn logout or after 14 days without use. Ornn verifies current membership for every request. The CLI does not accept an API key from a flag or environment variable. For unattended automation, call the REST API with an API key instead. See CLI sign-in and API keys.

Get help and discover commands

Use these three commands to launch, connect, and check compute:

ornn launch    Launch VM, Bare Metal, Kubernetes, or Slurm.
ornn connect   Open SSH, export Kubernetes cluster access, or enter Slurm access.
ornn check     Inspect lifecycle, health, metrics, and logs.

Start with bare ornn or scoped help:

ornn
ornn help launch
ornn help machines
ornn machines --help
ornn resources

Help is generated from the same command catalog the CLI executes. It shows the human form, explicit automation form, accepted references, review points, resumable states, and related next command.

Command	What it does
`ornn resources`	List the resource commands available to you.
`ornn status`	Check current Ornn platform availability.
`ornn version`	Show the installed version, machine-output schema, and update channel.
`ornn update [--major]`	Update within the installed major; `--major` acknowledges a major crossing.
`ornn completion <bash\|fish\|zsh> --format human`	Print shell completion for the command catalog.

ornn resources lists these command groups:

capacity-offers  list | show | terms | relist | delist
spot             books | book | trades | bid | orders | cancel
schedules        list | show | buy
offers           list | post | accept | cancel
reservations     list | show | invoices
machines         list | show | wait | switch | reboot | reset | metrics | logs
clusters         list | show | wait | eligible-machines | add-node | remove-node | teardown | metrics | logs
ssh-keys         list | add | delete
ssh-keys assignments  list | assign | sync | status
private-networks      list | show | create | update | delete
vpn-connections       list | configure | disconnect
volumes          list | show | create | refresh | clear | delete
volumes files    list | upload | download | delete
volumes folders  delete
volume-mounts    candidates | show | attach | wait | detach
external-buckets list | show | connect | update-credentials | verify | import | disconnect
filesystems      deploy | show | wait | delete

Use canonical IDs whenever possible. Every advertised exact, unique machine hostname works in human, JSON, and NDJSON output; the CLI resolves it to a canonical API request. Named resources are accepted only where scoped help lists them. The CLI never runs a fuzzy or first-result match.

Trade capacity

For listings with selectable terms and durations, make your offer in Reserve. CLI and MCP offer creation do not yet accept those selections.

ornn spot books
ornn spot book --gpu-slug GPU_SLUG
ornn spot bid --gpu-slug GPU_SLUG --gpus 8 --price-per-gpu-hour-micro-usd 2500000
ornn spot orders
ornn spot cancel --order-id ORDER_ID

ornn schedules list --listing-kind primary
ornn schedules show --listing-id LISTING_ID
ornn offers list
ornn offers cancel --listing-id LISTING_ID --offer-id OFFER_ID

ornn schedules list --listing-kind secondary
ornn schedules buy --listing-id LISTING_ID --starts-at START --ends-at END --ask-price-per-gpu-hour-micro-usd 4000000

ornn capacity-offers relist --reservation-id RESERVATION_ID --node-count 1 \
  --starts-at START --ends-at END \
  --terms '[{"minDurationHours":24,"maxDurationHours":48,"pricePerGpuHourMicroUsd":2500000,"renewable":false}]'
ornn capacity-offers list --kind secondary

Command	What it does
`ornn spot books`	List Spot order books by GPU type with their best bid and ask.
`ornn spot book --gpu-slug <slug>`	Show bid and ask depth for one GPU type.
`ornn spot trades --gpu-slug <slug>`	List recent Spot fills for one GPU type.
`ornn spot bid --gpu-slug <slug> --gpus <1\|2\|4\|8> [--order-type limit\|market] [--price-per-gpu-hour-micro-usd <n>]`	Place one Spot bid for a VM with 1, 2, 4, or 8 GPUs.
`ornn spot orders`	List your organization’s Spot bids and listing asks.
`ornn spot cancel --order-id <id>`	Cancel a resting bid or end a running Spot reservation.
`ornn schedules list [--listing-kind primary\|secondary] [--gpu-slug <slug>] [--from <time>] [--to <time>]`	List Reserve and Market listings with their open windows.
`ornn schedules show --listing-id <id>`	Show one listing with its sold windows and offers.
`ornn schedules buy --listing-id <id> --starts-at <time> --ends-at <time> --ask-price-per-gpu-hour-micro-usd <n>`	Buy a Market window at its asking price.
`ornn offers list`	List the offers your organization has posted.
`ornn offers post --listing-id <id> --price-per-gpu-hour-micro-usd <n> [--down-payment-cents <n>]`	Offer a price for a whole Reserve listing.
`ornn offers accept --listing-id <id> --offer-id <id>`	Accept one offer on a listing you own.
`ornn offers cancel --listing-id <id> --offer-id <id>`	Cancel one open offer.
`ornn capacity-offers list [--kind primary\|secondary\|on_demand] [--source-reservation-id <id>] [--limit <1-100>] [--cursor <opaque>]`	List capacity listings.
`ornn capacity-offers show --listing-id <id>`	Inspect one listing’s hardware, capacity, and availability.
`ornn capacity-offers terms --listing-id <id>`	Read a listing’s advertised prices and duration choices.
`ornn capacity-offers relist --reservation-id <id> --node-count <n> --starts-at <time> --ends-at <time> --terms <json> [--node-ids <ids>]`	Relist reserved capacity with explicit financial terms.
`ornn capacity-offers delist --listing-id <id>`	Delist a listing you relisted.

capacity-offers list and show return capacity and availability, including capacityListingId. Pass that ID to ornn capacity-offers terms --listing-id ID to read the advertised prices and duration choices separately. Each choice has a listingTermsId; a listing can have multiple choices. List pages accept --limit from 1 to 100 and an opaque --cursor from the previous response. spot places and cancels bids on the Spot order books. schedules lists Reserve (--listing-kind primary) and Market (--listing-kind secondary) listings; sold windows are removed. offers post, accept, and cancel offers on Reserve listings. schedules buy buys a window of a Market listing at its current asking price, paid up front. capacity-offers relist publishes capacity from a reservation with explicit financial terms; delist removes it from sale. Price editing is staff-only. Prices are integer micro-USD per GPU-hour (2500000 is $2.50), already include Ornn’s fees, and match the web console; times are RFC 3339. Every bid, offer, purchase, and relist asks for review first; add --yes in automation. Saving a card and paying invoices use the web app. See Spot, Reserve, the Market, and Relist capacity.

Reservations and invoices

ornn reservations list --kind schedule
ornn reservations show --reservation-id RESERVATION_ID
ornn reservations invoices --reservation-id RESERVATION_ID

Command	What it does
`ornn reservations list [--kind schedule\|spot] [--status <status>] [--listing-id <id>] [--cursor <opaque>]`	List your organization’s schedule and Spot reservations.
`ornn reservations show --reservation-id <id>`	Inspect one reservation.
`ornn reservations invoices --reservation-id <id>`	List one reservation’s invoices.

Launch compute

ornn launch RESERVATION_ID --mode vm
ornn launch RESERVATION_ID --mode bare-metal
ornn launch --mode kubernetes --reservation RESERVATION_ID --nodes 3
ornn launch --mode kubernetes --reservation RESERVATION_ID --nodes 0 --vms VM_ID,VM_ID
ornn launch --mode slurm --reservation RESERVATION_ID --key SAVED_KEY_LABEL --nodes 3

ornn launch [<reservation>] --mode vm|bare-metal|kubernetes|slurm [--key <ssh-key>] launches one reviewed VM, Bare Metal machine, Kubernetes cluster, or Slurm cluster from active reserved capacity. Kubernetes and Slurm prompt for a missing --network public|private and --nodes <count>. Bare Metal can add --network shared --volume <managed-volume> --mount-access read-only|read-write for a durable workspace. Launch uses active reserved capacity. In automation, VM, Bare Metal, and Slurm require --key <canonical-key-id>. Kubernetes and Slurm automation require one canonical organization, exact machine IDs in --machines matching --nodes, explicit --yes, and a stable idempotency key before any API request. --vms adds running VMs as workers, by canonical VM ID; --nodes 0 launches a cluster of VMs only. A VM’s SSH keys are set aside while it is in the cluster. Both modes recheck the organization and eligible machines immediately after review and fail closed if either changed. When a human composes a durable Bare Metal workspace with a private network and managed volume, the CLI generates and stores its recovery identity locally before the first mutation. An interrupted run prints ornn launch --mode bare-metal --resume; no workflow ID needs to be copied. Automation continues to supply its own stable --idempotency-key or --resume ID.

Connect

ornn connect
ornn connect MACHINE_ID_OR_EXACT_HOSTNAME
ornn connect MACHINE_ID_OR_EXACT_HOSTNAME --print
ornn connect CLUSTER_ID
ornn connect CLUSTER_ID --output ./ornn-cluster.yaml

ornn connect [<hostname-or-machine-id-or-cluster-id-or-reservation-id>] accepts --identity-file <path>, --user <login>, --output <path>, --print, and --overwrite. VM and Bare Metal use SSH. A direct connect announces the native command before opening it. Bare ornn connect discovers all current VM, Bare Metal, Kubernetes, and Slurm targets, then lets you choose among the combined set. Add --print to display a usable direct SSH command without executing it. For Kubernetes connection setup, follow Kubernetes access. Slurm uses SSH and native Slurm tools. Secret bytes are available only through an explicit export; raw --print cannot be combined with an explicit --format.

Check health, metrics, and logs

ornn check MACHINE_ID_OR_EXACT_HOSTNAME_OR_CLUSTER_ID
ornn machines metrics MACHINE_ID --start START --end END --max-points 240
ornn machines logs MACHINE_ID --stream kernel --limit 200

check combines lifecycle, health, reachability, metrics, and logs without hiding source failures, and human output includes returned metric/log samples.

Command	What it does
`ornn check [<id-or-hostname>] [--follow] [--start <time> --end <time>]`	Check one machine or cluster without treating missing evidence as healthy.
`ornn machines metrics [<machine>] [--start <time> --end <time>] [--max-points <1-500>] [--follow]`	Return one machine’s metrics for a window.
`ornn machines logs [<machine>] [--stream kernel\|serial] [--limit <1-500>] [--cursor <cursor>] [--follow]`	Return one machine’s logs with opaque pagination.
`ornn clusters metrics [<cluster>] [--start <time> --end <time>] [--max-points <1-500>] [--follow]`	Return per-machine metrics for a cluster.
`ornn clusters logs [<cluster>] [--stream kernel\|serial] [--limit <1-500>] [--follow]`	Return per-machine logs for a cluster.

Manage VM and Bare Metal machines

Command	What it does
`ornn machines list [--limit <1-500>] [--cursor <cursor>]`	List your VM and Bare Metal machines with lifecycle and SSH reachability.
`ornn machines show <machine>`	Show one machine, its lifecycle, and SSH reachability.
`ornn machines wait <machine> [--timeout <duration>]`	Wait for usable SSH. A timeout does not cancel remote work.
`ornn machines switch <reservation-id> [--mode vm\|bare-metal] [--network public\|private] [--erase-local-data]`	Change only the requested access mode or network, keeping SSH-key choices.
`ornn machines reboot <machine> [--no-wait]`	Reboot one machine and normally wait for SSH to return.
`ornn machines reset <machine> --erase-local-data [--no-wait]`	Hard reset one machine. Local data is erased.

<machine> is a canonical machine ID or an exact hostname.

Manage Kubernetes and Slurm clusters

Command	What it does
`ornn clusters list [--limit <1-500>] [--cursor <cursor>]`	List clusters without credentials.
`ornn clusters show [<cluster-id>]`	Show lifecycle and membership.
`ornn clusters wait [<cluster-id>] [--timeout <duration>]`	Wait for usable access with a bounded timeout.
`ornn clusters eligible-machines [<cluster-id>] [--limit <1-500>] [--cursor <cursor>]`	List the machine IDs and hostnames currently eligible to join.
`ornn clusters add-node [<cluster-id>] [<machine>] [--no-wait]`	Add one eligible machine.
`ornn clusters remove-node [<cluster-id>] [<machine>] [--no-wait]`	Remove one member.
`ornn clusters teardown [<cluster-id>] [--no-wait]`	Tear down the cluster. Reservation charges continue.

Launch clusters with ornn launch and get access with ornn connect.

Manage SSH keys

Command	What it does
`ornn ssh-keys list [--limit <1-500>] [--cursor <cursor>]`	List active organization SSH keys, 50 per page by default.
`ornn ssh-keys add --public-key-file <path> --label <label>`	Add one organization public key. `--public-key <key>` passes it inline.
`ornn ssh-keys delete <key-id-or-label>`	Revoke one organization key.
`ornn ssh-keys assignments list <reservation-id>`	List the keys assigned to one reservation.
`ornn ssh-keys assignments assign <reservation-id> --key <key-id-or-label>`	Assign one key to a reservation.
`ornn ssh-keys assignments sync <reservation-id>`	Queue delivery of the assigned keys to every machine.
`ornn ssh-keys assignments status <reservation-id>`	Show whether each assigned key is installed on each machine.

Assignment and delivery are separate. A queued key is not installed. See Manage SSH keys.

Private networks and VPN connections

Command	What it does
`ornn private-networks list [--limit <1-500>] [--cursor <cursor>]`	List your organization’s private networks.
`ornn private-networks show <name-or-id>`	Inspect one private network and its readiness.
`ornn private-networks create --name <name> [--cidr <cidr>]`	Create a private network.
`ornn private-networks update <name-or-id> --name <new-name>`	Rename a private network.
`ornn private-networks delete <name-or-id>`	Delete a private network when it can be deleted.
`ornn vpn-connections list <reservation-id> [--limit <1-500>] [--cursor <cursor>]`	List VPN connections without their configuration secrets.
`ornn vpn-connections configure <reservation-id> --name <name> --private-key <path> (--print \| --output <path>)`	Configure a VPN connection and export its configuration.
`ornn vpn-connections disconnect <reservation-id> <name-or-id>`	Disconnect one VPN connection without deleting the network.

Private networking belongs to the selected organization. Durable Bare Metal workspaces use --network shared with a managed volume. Per-reservation network attachment commands have been removed.

Storage

Command	What it does
`ornn volumes list [--limit <1-100>] [--cursor <cursor>]`	List managed volumes.
`ornn volumes show <volume>`	Inspect one volume.
`ornn volumes create --name <name> [--source <volume>]`	Create one durable volume.
`ornn volumes refresh <volume>`	Reconcile a volume’s usage with its stored files.
`ornn volumes clear <volume>`	Permanently clear a volume. Interactive terminal only.
`ornn volumes delete <volume>`	Delete one detached volume.
`ornn volumes files list <volume> [--prefix <path>] [--limit <1-500>] [--cursor <cursor>]`	List files by remote path.
`ornn volumes files upload <volume> <local-file> [--path <remote-path>] [--content-type <type>]`	Upload one file through a staged transfer.
`ornn volumes files download <volume> <remote-path> --output <local-path> [--overwrite]`	Download one file.
`ornn volumes files delete <volume> <remote-path>`	Permanently delete one file. Interactive terminal only.
`ornn volumes folders delete <volume> <remote-path>`	Permanently delete one folder tree. Interactive terminal only.
`ornn volume-mounts candidates [--limit <1-500>] [--cursor <cursor>]`	List Bare Metal volume-mount candidates.
`ornn volume-mounts show <reservation-id>`	Show per-machine mount state.
`ornn volume-mounts attach <reservation-id> --volume <volume> --access read-only\|read-write [--no-wait]`	Mount one volume on a reservation’s machines.
`ornn volume-mounts wait <reservation-id> [--timeout <duration>]`	Wait for every machine’s mount.
`ornn volume-mounts detach <reservation-id>`	Unmount the volume; the volume is kept.
`ornn filesystems deploy <reservation-id> --performance-tier <tier> --capacity-gib <gib> [--no-wait]`	Deploy a shared filesystem and mount it on every machine.
`ornn filesystems show <reservation-id>`	Show per-machine filesystem readiness.
`ornn filesystems wait <reservation-id> [--timeout <duration>]`	Wait for every filesystem member.
`ornn filesystems delete <reservation-id>`	Detach the filesystem; the underlying volume is kept.
`ornn external-buckets list [--limit <1-100>] [--cursor <cursor>]`	List external bucket connections without stored credentials.
`ornn external-buckets show <external-bucket>`	Inspect one connection.
`ornn external-buckets connect <gcs\|s3\|r2> --bucket <bucket> [options]`	Connect an external bucket. Verification is a separate step.
`ornn external-buckets update-credentials <external-bucket> --access-key-id <id> --secret-access-key-file <path>`	Replace a connection’s credentials.
`ornn external-buckets verify <external-bucket>`	Record whether Ornn can currently read the source.
`ornn external-buckets import <gcs\|s3\|r2> <external-bucket> --reservation <reservation-id>`	Copy the source into one managed working copy.
`ornn external-buckets disconnect <external-bucket>`	Remove one connection without writing to its source.

Storage list pages

Ornn CLI 1.0.22 and later request storage lists one page at a time. ornn volumes list and ornn external-buckets list accept --limit 1 through --limit 100 (default 25), ordered by last updated. Pass the returned page.next_cursor to --cursor for the next page; null means the end. Refresh from the first page after changes. Older CLI versions receive an upgrade error for storage lists. File-list pagination is unchanged.

Retry uploads and refresh usage

CLI 1.0.28 supports retrying a file upload with the same --idempotency-key and unchanged volume, remote path, file bytes, and content type. After the matching server update, a completed retry returns the original upload ID without sending bytes again. Use a new key for a different request or an expired incomplete upload. Overwriting a file replaces its contribution to volume usage. Run ornn volumes refresh <volume> to reconcile usage with the stored files.

GCS imports and Storage Transfer Service

CLI 1.x can connect a GCS bucket, but imports are currently blocked because GCS source verification is unavailable. Contact Ornn support to arrange an import. Include your source bucket name and region, destination reservation ID, and the files or prefix you need. Confirm the transfer cost, destination storage, and access permissions before starting. Ornn copies the source data into destination storage. Editing imported files does not update the originals in GCS. Your bucket administrator may need to grant read access; confirm the current transfer identity with Ornn before changing bucket permissions. Never share an access key or service-account private key. Wait for copying and mounting to finish, then check files from your workload. An accepted request does not mean the import is complete. Before replacing an existing attachment, export important data and confirm how to restore access. See storage pricing and the access guide for your mode.

CLI 0.2.x storage commands in historical release notes do not work in CLI 1.x. Use ornn help for the installed version’s commands.

Human and automation modes

In a terminal, commands default to human output. They may guide safe selection, show one final review, and ask default-no [y/N] consent. Pressing Enter, reaching EOF, or declining sends no mutation. Redirected and automated commands require an explicit format:

ornn status --format human
ornn status --format json
ornn machines list --limit 100 --format ndjson
ornn launch RESERVATION_ID --mode vm --yes \
  --idempotency-key WORKFLOW_ID --format json

The shared modifiers are --format human|json|ndjson, --guided, --yes, --wait, --no-wait, --timeout, --no-browser, --overwrite, and --idempotency-key. The old --json flag is not an alias; use --format json. Machine mode never prompts, launches a browser, or guesses a missing choice. Unsafe or incomplete input fails before any request. Mutations that create or increase spend, erase data, tear down a resource, or reduce access require explicit consent and a caller-stable idempotency key. Lists use bounded --limit and opaque --cursor values. Waits poll only transitional states, honor your timeout and cancellation, and report the last observed state on timeout. Use --no-wait to stop at the first durable state.

Read machine output

JSON and NDJSON use schema ornn.cli/v1:

{
  "schema": "ornn.cli/v1",
  "type": "result",
  "command": "machines show",
  "outcome": { "class": "ready", "phase": "ready" },
  "data": {},
  "action": null,
  "page": null,
  "errors": [],
  "resume": null
}

NDJSON emits progress records followed by exactly one terminal result or error.

Exit	Meaning
0	Requested goal reached
1	Failed, declined, or terminal negative result
2	Usage or validation error
3	Durable external action required
4	Useful partial result with named source errors
5	Sign-in or permission failure
6	Service unavailable or retryable failure
124	Timeout with last observed state
130	Interrupted

CLI and REST equivalents

Each command below maps to the REST request that does the same job with an API key. Paths are relative to https://api.ornn.com. Where a command composes several requests, the table lists each one. CLI-only flows have no documented REST route yet; use the CLI for those. The CLI adds review prompts, waits, and name resolution on top of these requests. Over REST, poll the read endpoints yourself and pass canonical IDs. See conventions for errors, pagination, and idempotency.

Session and platform

CLI command	REST request	Notes
`ornn login`	None	Create an API key in the console.
`ornn context`	`GET /v1/auth/session`	A key resolves to exactly one organization.
`ornn logout`	None	Revoke the key in the console.
`ornn status`	None	See status.ornn.com.
`ornn version`, `update`, `completion`, `resources`	None	Local to the CLI.

Capacity: Spot, Reserve, Market, and Relist

CLI command	REST request	Notes
`ornn spot books`	`GET /v1/markets/spot/books`
`ornn spot book`	`GET /v1/markets/spot/books/{gpuSlug}`
`ornn spot trades`	`GET /v1/markets/spot/books/{gpuSlug}/trades`
`ornn spot bid`	`POST /v1/markets/spot/orders`	Requires `Idempotency-Key`.
`ornn spot orders`	`GET /v1/markets/spot/orders`
`ornn spot cancel`	`DELETE /v1/markets/spot/orders/{orderId}`
`ornn schedules list`	`GET /v1/markets/schedules`
`ornn schedules show`	`GET /v1/markets/schedules/{listingId}`
`ornn schedules buy`	`POST /v1/markets/schedules/purchases`	Requires `Idempotency-Key`.
`ornn offers list`	`GET /v1/markets/offers`
`ornn offers post`	`POST /v1/markets/schedules/{listingId}/offers`	Requires `Idempotency-Key`.
`ornn offers accept`	`POST /v1/markets/schedules/{listingId}/offers/{offerId}/accept`
`ornn offers cancel`	`DELETE /v1/markets/schedules/{listingId}/offers/{offerId}`
`ornn capacity-offers list`	`GET /commerce/v2/listings`
`ornn capacity-offers show`	`GET /commerce/v2/listings/{listingId}`
`ornn capacity-offers terms`	`GET /commerce/v2/listings/{listingId}/terms`
`ornn capacity-offers relist`	`POST /commerce/reservations/{reservationId}/relist`	Requires `Idempotency-Key`.
`ornn capacity-offers delist`	`POST /commerce/v2/listings/{listingId}/delist`

Reservations and billing

CLI command	REST request	Notes
`ornn reservations list`	`GET /commerce/reservations`	Follow `nextCursor`.
`ornn reservations show`	`GET /commerce/reservations/{reservationId}`
`ornn reservations invoices`	`GET /commerce/reservations/{reservationId}/invoices`

VM and Bare Metal machines

CLI command	REST request	Notes
`ornn launch --mode vm` or `--mode bare-metal`	`POST /v1/orchestrator/nodes/reservations/{reservation_id}/launch`
`ornn launch --mode bare-metal --network shared --volume`	CLI only	The durable workspace also mounts a volume.
`ornn machines list`	`GET /v1/orchestrator/nodes?reservation_id={reservation_id}`	One request per reservation from `GET /commerce/reservations`.
`ornn machines show`, `machines wait`	`GET /v1/orchestrator/nodes/{node_id}`	`wait` polls this request.
`ornn machines switch`	`POST /v1/orchestrator/nodes/reservations/{reservation_id}/switch-access`
`ornn machines reboot`	`POST /v1/orchestrator/nodes/{node_id}/reboot`	Body `{"mode": "reboot"}`.
`ornn machines reset --erase-local-data`	`POST /v1/orchestrator/nodes/{node_id}/reboot`	Body `{"mode": "hard_reset"}`. Erases local data.
`ornn connect <machine>`	`GET /v1/orchestrator/nodes/{node_id}`	Read the connection details, then connect with SSH.

SSH keys

CLI command	REST request	Notes
`ornn ssh-keys list`	`GET /v1/organizations/{organization_id}/ssh-keys`	The CLI sends `status=active&limit=50`.
`ornn ssh-keys add`	`POST /v1/organizations/{organization_id}/ssh-keys`
`ornn ssh-keys delete`	`DELETE /v1/organizations/{organization_id}/ssh-keys/{key_id}`
`ornn ssh-keys assignments list`	`GET /v1/orchestrator/nodes/reservations/{reservation_id}/ssh-keys`
`ornn ssh-keys assignments assign`	`POST /v1/orchestrator/nodes/reservations/{reservation_id}/ssh-keys`
`ornn ssh-keys assignments sync`	`POST /v1/orchestrator/nodes/{node_id}/push-keys`	One request per machine.
`ornn ssh-keys assignments status`	`GET /v1/orchestrator/nodes/reservations/{reservation_id}/ssh-keys` and `GET /v1/orchestrator/nodes?reservation_id={reservation_id}`	Compare key fingerprints with each node’s `authorized_key_fingerprints`.
No CLI command	`DELETE /v1/orchestrator/nodes/reservations/{reservation_id}/ssh-keys/{key_id}`	Unassign a key from a reservation.

Kubernetes and Slurm clusters

CLI command	REST request	Notes
`ornn launch --mode kubernetes` or `--mode slurm`	`POST /v1/orchestrator/controllers`	For Slurm, `ssh_key_ids` selects login keys; omitted, every active organization key is used.
`ornn clusters list`, `show`, `wait`	`GET /v1/orchestrator/controllers`	Select by `controller.id`; `wait` polls this request.
`ornn clusters eligible-machines`	`GET /v1/orchestrator/nodes?reservation_id={reservation_id}`	Choose nodes that are not in a cluster (`controller_id` is null) and pass `health_readiness`; the add request rechecks eligibility.
`ornn clusters add-node`	`POST /v1/orchestrator/controllers/{controller_id}/nodes/{node_id}`	VM workers use `.../virtual-machines/{virtual_machine_id}`.
`ornn clusters remove-node`	`DELETE /v1/orchestrator/controllers/{controller_id}/nodes/{node_id}`	VM workers use `.../virtual-machines/{virtual_machine_id}`.
`ornn clusters teardown`	`POST /v1/orchestrator/controllers/{controller_id}/teardown`	Reservation charges continue.
`ornn connect <cluster>`	`GET /v1/orchestrator/controllers/{controller_id}/access`	Returns cluster credentials; store them privately.
`ornn clusters metrics`, `clusters logs`	See Monitoring	One request per worker node.

Private networks and VPN connections

CLI command	REST request	Notes
`ornn private-networks list`	`GET /v1/orchestrator/networks`
`ornn private-networks show`	`GET /v1/orchestrator/networks/{network_id}`
`ornn private-networks create`	`POST /v1/orchestrator/networks`
`ornn private-networks update`	`PATCH /v1/orchestrator/networks/{network_id}`
`ornn private-networks delete`	`DELETE /v1/orchestrator/networks/{network_id}`
`ornn vpn-connections list`	`GET /v1/orchestrator/networks/shared/peers`	Organization-wide; the request takes no reservation ID.
`ornn vpn-connections configure`	`POST /v1/orchestrator/networks/shared/peers`	Organization-wide; the request takes no reservation ID.
`ornn vpn-connections disconnect`	`DELETE /v1/orchestrator/networks/shared/peers/{peer_id}`

Storage

CLI command	REST request	Notes
`ornn volumes list`	`GET /v1/orchestrator/drives`
`ornn volumes show`	`GET /v1/orchestrator/drives/{drive_id}`
`ornn volumes create`	`POST /v1/orchestrator/drives`
`ornn volumes refresh`	`GET /v1/orchestrator/drives/{drive_id}?refresh=true`
`ornn volumes clear`	`DELETE /v1/orchestrator/drives/{drive_id}/contents`	Permanently deletes every file.
`ornn volumes delete`	`DELETE /v1/orchestrator/drives/{drive_id}`
`ornn volumes files list`	`GET /v1/orchestrator/drives/{drive_id}/contents`
`ornn volumes files upload`	`POST /v1/orchestrator/drives/{drive_id}/uploads`, then `POST /v1/orchestrator/drives/{drive_id}/uploads/{upload_id}`	Accepts `Idempotency-Key`. Upload the bytes between the two requests.
`ornn volumes files download`	`GET /v1/orchestrator/drives/{drive_id}/files/{file_path}`
`ornn volumes files delete`	`DELETE /v1/orchestrator/drives/{drive_id}/files/{file_path}`
`ornn volumes folders delete`	`DELETE /v1/orchestrator/drives/{drive_id}/folders/{folder_path}`
`ornn external-buckets list`, `show`	`GET /v1/orchestrator/drives`, `GET /v1/orchestrator/drives/{drive_id}`	External bucket connections are listed with volumes.
`ornn external-buckets connect`	`POST /v1/orchestrator/drives`
`ornn external-buckets update-credentials`	`PATCH /v1/orchestrator/drives/{drive_id}`
`ornn external-buckets verify`	`PATCH /v1/orchestrator/drives/{drive_id}`
`ornn external-buckets disconnect`	`DELETE /v1/orchestrator/drives/{drive_id}`
`ornn external-buckets import`	`PATCH /v1/orchestrator/drives/{drive_id}`	Send the destination `reservation_id`; see the storage reference for the deploy fields.
`ornn volume-mounts candidates`	`GET /v1/orchestrator/storage-targets`
`ornn volume-mounts attach`, `detach`	`PATCH /v1/orchestrator/drives/{drive_id}`	Send the `reservation_id`; body fields select attach or detach.
`ornn volume-mounts show`, `wait`	`PATCH /v1/orchestrator/drives/{drive_id}`	With only `reservation_id`, returns the volume and its `attachment`; `wait` polls it.
`ornn filesystems deploy`, `show`, `wait`, `delete`	CLI only

Monitoring

CLI command	REST request	Notes
`ornn check`	`GET /v1/orchestrator/nodes/{node_id}` or `GET /v1/orchestrator/controllers`, plus the metrics and log requests below	Combines lifecycle, metrics, and logs.
`ornn machines metrics`, `clusters metrics`	`GET /v1/observability/nodes/{node_id}/telemetry/latest`, `GET /v1/observability/nodes/{node_id}/telemetry/query`	Latest sample, or a window with `start`, `end`, and `max_points`.
`ornn machines logs`, `clusters logs`	`GET /v1/observability/nodes/{node_id}/logs/{stream}/query`	`stream` is `kernel` or `serial`.

Migrating to vNext

The current major is a clean cutover. A retired command does not execute. It exits before making a request and prints one exact replacement plus the migration URL.

Previous form	Use now
`ornn account`, `ornn whoami`	`ornn context`
`catalog search`, `gpu-types`, `availability `, `listing `	`capacity-offers list` or `show`; `schedules list`
`ornn buy`	`offers post`
`bid create`, `bid update`	`spot bid`
`bid list`, `bid show`	`spot orders`
`bid withdraw`	`spot cancel`
`exchange *`, `gpu` or `gpus` list, show, checkout	`capacity-offers list`
`reservations checkout`	`reservations show`
`ornn nodes launch`	`ornn launch`
`nodes` list, show, wait, switch, reboot, hard-reset	matching `machines` leaf; reset needs `--erase-local-data`
`ssh`, `nodes ssh-command`	`connect`
`metrics`, `telemetry`, `logs`	`check`, `machines metrics`, or `machines logs`
`clusters create`, `kubernetes launch`, `slurm launch`	`launch --mode kubernetes` or `launch --mode slurm`
Kubernetes or Slurm status, credentials, SSH, teardown	`clusters show`, `connect`, or `clusters teardown`
`network `, `networks `	`private-networks <verb>`
`access *`, `keys`, `nodes keys`	the `machines` or `ssh-keys` replacement printed by the CLI
`storage volumes *`	`volumes <verb>`
`storage files *`	`volumes files <verb>`
`storage buckets *`	`external-buckets <verb>`
Storage targets, deploy, status, detach, filesystem	the `volume-mounts` or `filesystems` replacement printed
`billing`	Use the Ornn web app.
`api`, `cache purge`	`resources`; arbitrary API dispatch is removed
`--json`	`--format json`
`--interactive`	`--guided`

The new contract also uses deadline-bound waits, semantic default-no consent, bounded cursor pagination, versioned machine envelopes, and semantic exit statuses. @ornncompute/cli 1.x uses the existing ornn command and requires Node 20 or newer. Install the latest published release with npm install -g @ornncompute/cli@latest. Publication does not wait for the legacy support schedule, which starts at 2026-10-05T00:00:00Z. 0.2.10 remains the configured previous release for rollback. ornn update updates within the installed major. Review this guide, then use ornn update --major to acknowledge the major crossing. Through day 90, you can roll back to the final previous release:

npm install -g @ornncompute/cli@0.2.10

The new and previous releases reuse valid auth.json, config.json, endpoint settings, and update cache. Updating or rolling back does not rewrite your bearer. Rollback is unsupported at 2027-01-04T00:00:00Z (day 91). Legacy calls then return an upgrade-only 426 response through day 180. The retired routes are absent at 2027-04-04T00:00:00Z (day 181). Use ornn --version or ornn -v for the semantic version. ornn version also shows the machine schema and update channel.

Get Started

Buy and sell

Supplier

Accessing Compute

Storage

Nodes

Account

Install

Get help and discover commands

Trade capacity

Reservations and invoices

Launch compute

Connect

Check health, metrics, and logs

Manage VM and Bare Metal machines

Manage Kubernetes and Slurm clusters

Manage SSH keys

Private networks and VPN connections

Storage

Storage list pages

Retry uploads and refresh usage

GCS imports and Storage Transfer Service

Human and automation modes

Read machine output

CLI and REST equivalents

Session and platform

Capacity: Spot, Reserve, Market, and Relist

Reservations and billing

VM and Bare Metal machines

SSH keys

Kubernetes and Slurm clusters

Private networks and VPN connections

Storage

Monitoring

Migrating to vNext

​Install

​Sign in and choose an organization

​Get help and discover commands

​Trade capacity

​Reservations and invoices

​Launch compute

​Connect

​Check health, metrics, and logs

​Manage VM and Bare Metal machines

​Manage Kubernetes and Slurm clusters

​Manage SSH keys

​Private networks and VPN connections

​Storage

​Storage list pages

​Retry uploads and refresh usage

​GCS imports and Storage Transfer Service

​Human and automation modes

​Read machine output

​CLI and REST equivalents

​Session and platform

​Capacity: Spot, Reserve, Market, and Relist

​Reservations and billing

​VM and Bare Metal machines

​SSH keys

​Kubernetes and Slurm clusters

​Private networks and VPN connections

​Storage

​Monitoring

​Migrating to vNext

Install

Sign in and choose an organization

Get help and discover commands

Trade capacity

Reservations and invoices

Launch compute

Connect

Check health, metrics, and logs

Manage VM and Bare Metal machines

Manage Kubernetes and Slurm clusters

Manage SSH keys

Private networks and VPN connections

Storage

Storage list pages

Retry uploads and refresh usage

GCS imports and Storage Transfer Service

Human and automation modes

Read machine output

CLI and REST equivalents

Session and platform

Capacity: Spot, Reserve, Market, and Relist

Reservations and billing

VM and Bare Metal machines

SSH keys

Kubernetes and Slurm clusters

Private networks and VPN connections

Storage

Monitoring

Migrating to vNext