Server setup

Configure only the app credentials.

Vela Console issues the App ID and App token. Execution scope is resolved from app or Console state, not from required server environment variables.

Environment variables

Keep required server environment to the credentials that let the SDK talk to Vela. Customer, delegation, and model selection belong in app-owned settings, Hosted Connect return state, or Vela Console-managed integration state.

Vela environment variables used by the SDK
Variable Requirement Purpose
VELA_APP_ID Required Vela Console App ID. The SDK fills this into app API calls after client creation.
VELA_APP_TOKEN Required Server-side App token. Pass it as SDK appToken for manual clients; the legacy appKey alias is also accepted. Never expose it to the browser.
VELA_BASE_URL Optional override Only needed for local or preview Vela API development. Production defaults to https://api.vel4.ai.

Use getVelaEnvironmentDiagnostics({ env: process.env }) before creating the client so setup responses can return missing App credentials at once. The default diagnostics check treats VELA_BASE_URL as optional and does not require execution scope env vars.

Model execution still needs an execution scope: customerTenantId, delegationId, and optional requestedModel. Resolve those from the signed-in workspace, a saved Vela integration record, the Hosted Connect return URL, or Vela Console state. Do not make VELA_CUSTOMER_TENANT_ID, VELA_DELEGATION_ID, or VELA_REQUESTED_MODEL required application env vars for the MVP.