Standalone Select AI A2A deployment

gcloud/standalone/deploy.sh builds or selects a Select AI container image, creates or updates a private Cloud Run service, and configures its database secrets. Run it on a machine with the Google Cloud CLI installed and authenticated to the target project.

IAM permissions

The scripts use the active gcloud identity. They do not elevate its access.

Deployer (the active gcloud identity)

Operation	Required permissions
Inspect and create the Artifact Registry repository	artifactregistry.repositories.get, artifactregistry.repositories.create
Inspect and create the default runtime service account	iam.serviceAccounts.get, iam.serviceAccounts.create
Deploy or update Cloud Run	run.services.create, run.services.update, run.services.get, run.operations.get; iam.serviceAccounts.actAs on the runtime service account; artifactregistry.repositories.downloadArtifacts on the image repository
With --build, upload local source, submit, and wait for a build	storage.buckets.get, storage.objects.create on the configured source-staging bucket; cloudbuild.builds.create, cloudbuild.builds.get, serviceusage.services.use
Inspect, create, and add versions to database or wallet secrets	secretmanager.secrets.get, secretmanager.secrets.create, secretmanager.versions.add
Grant the runtime account access to those secrets	secretmanager.secrets.getIamPolicy, secretmanager.secrets.setIamPolicy
Grant Gemini Enterprise and the active gcloud identity access to the service	run.services.getIamPolicy, run.services.setIamPolicy
Obtain the project number	resourcemanager.projects.get

Runtime service account

Operation	Required permissions
Read database and wallet secrets while serving requests	secretmanager.versions.access

Other service identities

Principal	Operation	Required permissions
Cloud Build execution service account	With --build, push the built image	artifactregistry.repositories.uploadArtifacts
Gemini Enterprise service agent	Invoke the private Cloud Run service	run.routes.invoke
Active gcloud identity	Fetch the Agent Card after deployment	run.routes.invoke

The source-staging bucket is Cloud Build's default unless a custom bucket is configured. Cloud Build also needs access to its build-log destination; the default same-project build account has that access. If your organization uses a custom build service account, source bucket, or log bucket, its administrator must grant the equivalent Cloud Storage permissions on those resources.

Google Cloud references: Service Usage access control, Cloud Run deployment permissions, Secret Manager access control, Artifact Registry roles, and Cloud Build roles.

The rows that set IAM policy are administrative mutations. They are present because deploy.sh creates and rotates secrets and configures private-service invocation. If your customer deployment identity must not change IAM, provision the secrets and the secretmanager.versions.access/run.routes.invoke permissions beforehand, then remove those policy-setting commands from the deployment workflow.

Prerequisite: enable project APIs once

An administrator must enable these APIs once for the project:

gcloud services enable \
  run.googleapis.com \
  cloudbuild.googleapis.com \
  artifactregistry.googleapis.com \
  secretmanager.googleapis.com \
  discoveryengine.googleapis.com \
  --project PROJECT_ID

Deploy dynamic standalone

gcloud/standalone/deploy.sh \
  --connection-mode dynamic \
  --build

With no database options, the deployment fixes neither the Connection URL nor the AI Agent. The generated A2UI form asks each A2A conversation for Connection URL, username, password, and AI Agent. The resulting database session runs in an isolated child process and remains bound to the conversation until it expires. With --require-oauth, that binding also includes the authenticated owner. Dynamic standalone is restricted to one Cloud Run instance because its session routing is in memory.

Use --db-dsn-secret NAME to fix the Connection URL from an existing Secret Manager secret, and --a2a-team TEAM to fix the AI Agent. The generated form contains only the values that are not fixed. Use --rotate-db-config to prompt for a Connection URL and create or rotate the service's default DSN secret.

Dynamic deployment options	Generated A2UI fields
Neither option	Connection URL, Database username, Database password, AI Agent
--db-dsn-secret only	Database username, Database password, AI Agent
--a2a-team only	Connection URL, Database username, Database password
Both options	Database username, Database password

Fix the Connection URL and AI Agent

The following command updates the default dynamic service so every A2A conversation supplies only its database username and password. If the named secret does not exist, the script prompts for the Connection URL and creates it. Later invocations reuse the secret without prompting.

gcloud/standalone/deploy.sh \
  --project PROJECT_ID \
  --connection-mode dynamic \
  --a2a-team ORACLE_AI_DATABASE_AGENT \
  --db-dsn-secret select-ai-a2a-standalone-dynamic-db-connect-string

Add --rotate-db-config only when the secret already exists and its Connection URL must be replaced.

Choose whether end-user OAuth is required

By default, the service does not require an application OAuth token. Cloud Run IAM still allows only authorized callers such as Gemini Enterprise to invoke the private service. Each A2A conversation gets a separate database session, but the server does not know or verify which human user owns that conversation.

To require verified end-user ownership, add --require-oauth:

gcloud/standalone/deploy.sh \
  --connection-mode dynamic \
  --require-oauth \
  --build

In this mode every A2A request must include Authorization: Bearer ... and the session is scoped by both the authenticated owner and the conversation. When registering the agent, configure Gemini Enterprise end-user OAuth rather than selecting Skip & Finish. Gemini Enterprise automatically sends its Cloud Run invocation identity in X-Serverless-Authorization; end-user OAuth adds the separate Authorization header consumed by a2a serve.

The server derives a stable owner from an ID token's iss and sub claims. It also accepts opaque OAuth access tokens without storing or logging them, but a refreshed opaque token starts a new owner/session scope. See Register and manage A2A agents for the OAuth registration steps.

Deploy a separate OAuth-enabled service from an existing image

A new Cloud Run service cannot inherit an image from another service. Resolve the most recently updated Select AI image directly from Artifact Registry and construct an immutable digest URI:

PROJECT_ID=PROJECT_ID
REGION=us-central1
REPOSITORY=select-ai
IMAGE_NAME=select-ai

IMAGE_RECORD="$(gcloud artifacts docker images list \
  "$REGION-docker.pkg.dev/$PROJECT_ID/$REPOSITORY/$IMAGE_NAME" \
  --project "$PROJECT_ID" \
  --include-tags \
  --sort-by='~UPDATE_TIME' \
  --limit=1 \
  --format='csv[no-heading](package,version)')"

if [[ -z "$IMAGE_RECORD" ]]; then
  echo "No Select AI image exists in Artifact Registry." >&2
  exit 1
fi

IMAGE_URI="${IMAGE_RECORD/,/@}"
echo "$IMAGE_URI"

Then create a separate service that fixes the Connection URL and AI Agent and requires end-user OAuth:

gcloud/standalone/deploy.sh \
  --project "$PROJECT_ID" \
  --region "$REGION" \
  --repository "$REPOSITORY" \
  --connection-mode dynamic \
  --service select-ai-a2a-standalone-oauth \
  --a2a-team ORACLE_AI_DATABASE_AGENT \
  --db-dsn-secret select-ai-a2a-standalone-oauth-db-connect-string \
  --require-oauth \
  --image-uri "$IMAGE_URI"

On its first invocation, the deployment script prompts for the Connection URL and creates the service-specific secret. The generated A2UI form contains only database username and password. Register this service in Gemini Enterprise with end-user OAuth; selecting Skip & Finish causes its requests to fail with HTTP 401.

Deploy fixed standalone

gcloud/standalone/deploy.sh \
  --connection-mode fixed \
  --a2a-team ORACLE_AI_DATABASE_AGENT \
  --build

Fixed mode prompts for the ADB connect descriptor, username, and password. It mounts all three from Secret Manager and creates the existing shared database pool at server startup. The generated A2UI connection form is skipped.

Optional: Autonomous Database mTLS wallet

The Select AI SDK already supports wallet_location and wallet_password. Wallet configuration is supported only in fixed mode. For Cloud Run, pass the path to the downloaded Autonomous Database wallet ZIP on the first deployment (or when replacing it):

gcloud/standalone/deploy.sh \
  --connection-mode fixed \
  --a2a-team ORACLE_AI_DATABASE_AGENT \
  --wallet-archive /path/to/Wallet_database.zip

The script prompts for the wallet password, stores the ZIP and password as service-specific Secret Manager secrets, and grants access only to the runtime service account. Cloud Run mounts the ZIP read-only; its A2A launcher expands it into ephemeral /tmp storage before starting the SDK, verifies it contains ewallet.pem, and sets SELECT_AI_WALLET_LOCATION to that file's directory. Do not commit the wallet ZIP or put its contents in the image.

Later deploys reuse the wallet. To replace it, pass --wallet-archive again.

The first deployment needs --build (or an explicit --image-uri). Later deployments reuse the image already deployed to the service, so changing Cloud Run configuration or secrets does not create another image. The command deploys private Cloud Run, sets the final public URL in the Agent Card, grants your active gcloud identity and Gemini Enterprise Discovery Engine service agent the run.routes.invoke permission for this Cloud Run service.

The default Cloud Run services are select-ai-a2a-standalone-dynamic and select-ai-a2a-standalone-fixed, selected by --connection-mode. Dynamic mode has no default AI Agent. Fixed mode requires --a2a-team so all four connection values are complete at startup. Fix either optional dynamic value with explicit options:

gcloud/standalone/deploy.sh \
  --connection-mode dynamic \
  --service sales-analyst-a2a \
  --a2a-team SALES_ANALYST

Use a distinct --service value for each fixed deployment profile. Each service gets distinct Secret Manager secret names by default, so credentials remain attached to that A2A server.

--max-instances controls the number of Cloud Run containers in fixed mode. Dynamic mode requires exactly one instance. A fixed-mode container can use up to 10 Oracle connections by default; change that limit with --pool-max-size.

Update the Select AI SDK or this repository

Update the checkout (or modify its dependency version), then explicitly build and deploy the new image:

git pull
gcloud/standalone/deploy.sh --connection-mode dynamic --build

--build creates a freshly tagged image from the current source; without it, the existing image is reused. Existing explicitly selected database secrets are reused without prompting. To configure or rotate the dynamic deployment's Connection URL, explicitly request it:

gcloud/standalone/deploy.sh \
  --connection-mode dynamic \
  --rotate-db-config

What cloudbuild.yaml does

gcloud/standalone/deploy.sh --connection-mode MODE --build uses gcloud/standalone/cloudbuild.yaml to tell Cloud Build to build docker/Dockerfile and push it to Artifact Registry. It is build configuration, not a command you run. The build context is the repository root, so the image can install the Select AI source from pyproject.toml and src/.

Cloud Build upload contents

Before the build starts, gcloud builds submit archives and uploads the repository root. The root .gcloudignore excludes local virtual environments, generated documentation, test data, caches, credentials, and Git metadata. Keep src/, pyproject.toml, docker/, and gcloud/ in the upload; they are required to build the image. If the upload is unexpectedly large, check local directories against .gcloudignore before running --build again.

After a successful deployment, the script prints the A2A Agent Card JSON. Paste that JSON into Gemini Enterprise to register the private service. The required Gemini Enterprise invocation permission has already been added.