Glossary
This glossary defines the Alien terms used across the docs, the CLI, and the dashboard, each in a sentence or two with a link to the page that explains it.
Architecture
Remote environment
Where a deployment runs: a customer's AWS account, Google Cloud project, or Azure subscription, a dedicated account in your own cloud, a Kubernetes cluster, or Linux servers on-prem. See How Alien works.
Control plane
In the most common way to use Alien, the part of your product you keep hosted in your own cloud, such as sign-in, billing, UI, and metadata, which calls into each customer's deployment. See Control plane ⇄ Data plane.
Data plane
In the same pattern, the smallest piece of your product that has to run next to the customer's data, private systems, or hardware, deployed into a remote environment. See Control plane ⇄ Data plane and Architecture design.
Manager
The Alien service that stores your releases and deployments and keeps each deployment on the right release. It updates cloud deployments through the customer's cloud APIs (push), and answers the Operators that connect to it from other environments (pull).
alien.dev runs managers for you, or you can run a Private Manager. See How Alien works.
Private Manager
A manager that runs in your own cloud account instead of alien.dev's, created under Settings → Private Managers or with alien managers create. See Private managers.
Operator
A small Alien process, alien-operator, that runs inside a pull-mode environment such as a Kubernetes cluster or Linux servers. It connects out to the manager over HTTPS, applies releases locally, and sends status and logs back.
Not to be confused with Remote Operator. See Push, pull, and air-gapped.
Building
Stack
The set of resources each deployment gets, declared in alien.ts with new alien.Stack(...). See Stacks.
Resource
One building block in a stack, such as a Worker, Container, Storage bucket, Kv table, or Queue. See Resources.
Link
A connection from a compute resource to another resource with .link(...), which gives the workload a binding and the permissions to use it. See Resources and links.
Binding
The typed client your code gets for a linked resource, such as storage("files") in the TypeScript SDK.
Frozen resource
A resource the customer's admin creates during initial setup with their own credentials, usually one that holds data, such as a bucket or a database. Releases can't replace or delete it, so the management identity needs fewer permissions. See Frozen and live resources.
Live resource
A resource Alien creates, updates, and replaces with the management identity on every release during ongoing management, such as a Worker or a Container. See Frozen and live resources.
Input
A value the stack asks for at initial setup, provided by you (the developer) or by the customer's admin (the deployer), such as an API key or a feature flag. See Inputs and secrets.
Permission profile
A named set of grants in alien.ts, such as execution with storage/data-read on one bucket. Each workload runs as one profile, and each profile becomes one cloud identity: an IAM role, a service account, or a managed identity. See Permissions.
Command
A named handler in your Worker, Container, or Daemon that your backend invokes with alien commands invoke or the API, without opening inbound ports. See Commands.
Remote binding
A binding your own backend uses to reach a resource inside a customer's environment, for resources added with { remoteAccess: true }. See Remote bindings.
Releasing
Release
An immutable build of your stack, with images pushed and a version recorded, created by alien release. See Releases.
Release channel
A named pointer to a release, such as production or staging, that deployments follow. alien release moves production to the new release, and promoting a release to a channel updates every deployment on it. See Releases and Rollouts.
Pin
A setting that fixes one deployment to a specific release so it stops following its release channel. See Rollouts.
Versionless release
A release that records only a version, created with alien release --no-stack --version <v>, for Remote Operator environments running software Alien didn't ship. See Releases.
Deploying
Deployment
One running copy of your stack in one remote environment. See Deployments.
Deployment group
Alien's record of one customer, or of one of your own environments. It holds that customer's deployments and an optional external ID. See Deployments.
External ID
Your own ID for a customer, such as your tenant or organization ID, stored on their deployment group. Use it to find the customer without storing Alien IDs.
The hosted gateways take it in the X-Alien-External-ID header. See REST API.
Push
The deployment model where the manager updates the environment through cloud APIs with the management identity. The default on AWS, Google Cloud, and Azure. See Push, pull, and air-gapped.
Pull
The deployment model where an Operator inside the environment connects out to the manager and deploys locally, with no cross-account access. Required on Kubernetes and Linux servers. See Push, pull, and air-gapped.
Air-gapped
A deployment with no network path to the manager, updated by carrying signed bundles in with alien-deploy sync. See Push, pull, and air-gapped.
Machine
A Linux host, on-prem or bare metal, that joins a Linux servers deployment (platform machines) and runs its workloads. See Linux servers.
Join token
A credential in the host join command that lets a machine join a deployment. You can list, revoke, and rotate join tokens. See Linux servers.
Capacity group
A named pool of machines in a deployment that workloads are scheduled onto. A host picks one with alien-deploy join --capacity-group. See Linux servers.
Deployment UX
Initial setup
The one-time step the customer's admin runs in their own cloud with their own credentials. It creates the management identity and the frozen resources, then registers the deployment with Alien.
Everything after it is ongoing management. See Deployment UX and Frozen and live.
Ongoing management
Everything Alien does after initial setup, without the customer's admin: it rolls out releases, keeps live resources current, reports status and logs, and runs commands. It runs through the management identity (push) or the Operator (pull), never the admin's credentials. See The two phases and Frozen and live.
Setup link
A URL for one customer that opens the deployment portal, where their admin runs the initial setup. Its token can only set up deployments in that customer's deployment group.
Create one with alien onboard, in the dashboard, or through the API, which returns it as deploymentLink. See Send a link.
Setup item
One thing a setup link sets up: the application, AI models, customer-managed keys, a storage bucket, a container registry, or a sandbox. See Setup items.
Deployment portal
The white-labeled setup page Alien hosts for the customer's admin, on your own domain. A setup link opens it. The admin picks a cloud and a setup method, gets the exact template, chart, or command to run, and follows the setup. The admin needs no Alien account.
The portal appears in two of the Deployment UX options:
- Send a link: you create a setup link and send it to the admin. No code.
- Launch the portal from your product: a Deploy button in your product creates the link and opens the portal, and your product shows status and removal.
The third, Build it into your product, replaces the portal with your own setup screen.
See Deployment portal.
Deploy button
The Deployment UX option Launch the portal from your product: a Deploy button in your own product. A click makes your backend create a setup link and opens the deployment portal; your page then shows the setup's progress. See Add a Deploy button.
Your own setup screen
The Deployment UX option Build it into your product: a setup page you build in your own product. Your backend creates the deployment, and the API returns the setup step to show: a CloudFormation launch link, or the Terraform, Helm, or CLI commands the admin runs. Google sign-in and Linux servers still go through the deployment portal. See Build your own setup screen.
Setup package
What Alien builds from each release for the customer's admin to run: a CloudFormation template, Terraform modules, a Helm chart with its Operator image, or a CLI. The dashboard lists them under Artifacts, and the API calls them packages. See Setup packages.
Management identity
The scoped identity initial setup creates in the customer's cloud so the manager can deploy updates: a cross-account IAM role on AWS (the management role), a service account on Google Cloud, or a managed identity on Azure.
Its permissions are derived from the stack's live resources and features. See Permissions.
alien-deploy
The CLI a customer's admin runs for initial setup and to manage a deployment, which you can white-label under your product name. See alien-deploy CLI.
Operating
Remote Operator
The part of Alien that lets you operate software in a remote environment through named operations and time-boxed access, built into every deployment or added alone to a product you already ship. See Remote Operator.
Operation
A named, reviewed action from the operations catalog, such as reading pod logs or restarting a workload, tagged Read-only, Mutating, or Destructive. See Operations.
Access request
A time-boxed grant (default and maximum 6 hours) to run operations or a debug session in one deployment, which can require approval. See Access requests.
Debug session
A local shell or command that runs with short-lived, scoped credentials for a deployment, started with alien debug. See Debug.
Heartbeat
The periodic health report a deployment's resources send to the manager. See Deployments.
Product features
Customer-managed keys
Encryption keys in the customer's own key service that the customer can revoke, also called bring your own key (BYOK). Resources in a deployment use them natively; a hosted app uses them through the Encrypt/Decrypt API. See Customer-managed keys.
Local AI models
Your product's AI calls run on the customer's own model account, also called bring your own LLM. A hosted app sends them through the AI Gateway; a deployment calls the customer's models directly. See Local AI models.
Encrypt/Decrypt API
The hosted endpoint (https://encryption.alien.dev) that encrypts and decrypts with a customer's own key for an app you host. Formerly called the Encryption Gateway. See Customer-managed keys.
AI Gateway
A hosted, OpenAI-compatible endpoint (https://ai.alien.dev/v1) that sends a hosted app's model calls to the customer's own Amazon Bedrock, Vertex AI, or Azure AI Foundry account. See Local AI models.
Remote Sandbox
An isolated environment for running untrusted code, such as an AI agent's tool calls, inside the customer's cloud. See Remote sandboxes.
Account
Workspace
The top-level account in the Alien dashboard that holds members, billing, and projects. Its name is permanent and appears in dashboard URLs and CLI commands: 4 to 100 lowercase letters, numbers, and hyphens.
Project
One product you deploy with Alien, with its own releases, customers, API keys, and settings.
API key
A secret starting with ax_, scoped to a workspace, project, deployment, or deployment group, used to call the API and CLI. See REST API.