Deployment and Cloud Providers

Deployment and Cloud Providers

NEXUS AI uses one deployment model across the dashboard, CLI, REST API, and MCP. A deployment can start from source code, an existing image, an AI Builder snapshot, or a GitHub binding.

Deployment inputs

Git source

NEXUS AI clones the repository and branch, detects or accepts build settings, builds a container image, and deploys it.

nexus deploy source \
  --repo https://github.com/you/app.git \
  --branch main \
  --name app \
  --provider docker \
  --wait

Private repositories use a stored secret through --repo-secret; do not place a token in the repository URL.

Container image

nexus deploy create \
  --image nginx:latest \
  --port 80 \
  --name web \
  --provider docker

The platform validates allowed images and Dockerfile behavior before deployment.

Local source and AI-generated source

The dashboard supports uploaded local source, one-shot prompt deployment, and AI Builder deployment. All use the same build and runtime lifecycle after source is accepted.

Provider matrix

Provider value	Runtime	Typical use
docker	NEXUS AI container runtime	Full-stack apps, sidecar databases, Redis, workers, volumes, and buckets
gcp_cloud_run	Google Cloud Run	Managed single-container app; optional supported managed DB
aws_ecs_fargate	AWS App Runner implementation	Managed single-container app; optional supported managed DB
azure_container_apps	Azure Container Apps	Managed single-container app; optional supported managed DB

The AWS CLI identifier is retained for compatibility even though the current deployment implementation targets App Runner.

Provider availability and allowed regions depend on plan and organization configuration.

NEXUS AI container runtime

Use docker when the application needs a locally orchestrated full stack:

• PostgreSQL, MySQL, MongoDB, or Redis sidecars;
• a background worker process;
• persistent filesystem volumes;
• S3-compatible object-storage buckets;
• container-level exec and single-file cp operations;
• service-to-service networking through internal hostnames.

Example:

nexus deploy source \
  --repo https://github.com/you/app.git \
  --name app \
  --provider docker \
  --services postgresql,redis \
  --worker-command "npm run worker" \
  --wait

Cloud provider targets

Cloud Run, App Runner, and Container Apps are single-container runtimes. Compose-style sidecars and locally mounted NEXUS AI volumes are not available there.

For a cloud deployment that needs a database, create or attach a supported managed database:

nexus deploy source \
  --repo https://github.com/you/app.git \
  --name app \
  --provider gcp_cloud_run \
  --region us-central1 \
  --create-db postgres \
  --wait

Or attach an existing managed database:

nexus deploy source \
  --repo https://github.com/you/app.git \
  --name app \
  --provider gcp_cloud_run \
  --managed-db <database-id> \
  --wait

Supported managed offerings include NEXUS AI local databases plus provider-backed AWS RDS, Google Cloud SQL, and Azure database services where configured.

Build configuration

Framework detection covers common Node.js, Python, Go, Ruby, PHP, Java, and static-web patterns. Override detection with:

• --framework;
• --install-command;
• --build-command;
• --start-command;
• --output-dir;
• --dockerfile.

--dockerfile accepts a repository-relative path, local file, URL, or inline content. Keep the build context small with .dockerignore.

Ports and health

• Bind the process to 0.0.0.0.
• Use the platform PORT value or ensure the configured service port matches.
• A successful build does not guarantee a healthy runtime.
• Long migrations or blocking startup work can cause health timeouts.
• Use --no-health-check only when the workload intentionally has no HTTP service.

nexus deploy status app
nexus deploy logs app --type build --lines 200
nexus deploy logs app --follow

Environments and configuration

Deployments use DEVELOPMENT, STAGING, or PRODUCTION. Use separate projects or deployment names for independently managed environments.

nexus deploy source \
  --repo https://github.com/you/app.git \
  --project <production-project-id> \
  --environment PRODUCTION \
  --env-file .env.production \
  --wait

Do not commit production secrets to an env file. Use the Secrets Vault for long-lived credentials.

Lifecycle operations

nexus deploy get app
nexus deploy status app --watch
nexus deploy logs app --follow
nexus deploy scale app 3
nexus deploy stop app
nexus deploy start app
nexus deploy redeploy app --wait
nexus deploy rollback app
nexus deploy delete app --yes

Stop and start preserve deployment configuration. Delete removes the deployment and can affect provider resources. Database, volume, and bucket lifecycle rules are separate; inspect attachments and backups before deleting a production stack.

Auto-destroy previews

Create an ephemeral environment:

nexus deploy source \
  --repo https://github.com/you/app.git \
  --branch feature/demo \
  --auto-destroy 4 \
  --wait

Change or remove the schedule with nexus deploy auto-destroy; check the installed command help for accepted duration and timestamp forms.

Custom domains

nexus domain add app api.example.com
nexus domain list app
nexus domain verify app <domain-id>
nexus domain remove app <domain-id>

Follow the DNS values returned by the platform. Keep proxying disabled while a provider requires direct ownership or TLS verification. DNS and certificate issuance can remain pending after the application itself is healthy.

Container access

nexus exec and nexus cp are available only for running docker deployments:

nexus exec <deployment-id> -- ls -la /app
nexus cp ./index.html <deployment-id>:/app/public/index.html
nexus cp <deployment-id>:/app/config.json ./config.json

Copied files change the running container, not the immutable release. They can disappear after redeploy or rollback; commit durable fixes to source.

Related pages

• Quick Start
• GitHub Deployments
• Databases, Backups, and Storage
• Versioning and Rollbacks
• CLI Reference