Configure your first sync

Build and validate isolated Google Ads and Meta Ads connection definitions.

This tutorial is for a developer composing the public package into an operator-owned Cloudflare deployment. You will create typed desired state and turn it into validated connection definitions without putting credential values in source.

define desired state

const config = defineAdsSyncConfig({
  connections: [
    {
      catalog: googleCatalog,
      connectionId: "google_ads_performance",
      destinationSchema: "airbyte_google_ads_performance",
      provider: "google_ads",
      selectedStreams: ["campaign_daily_performance"],
      sourceConfigSecret: "GOOGLE_ADS_SOURCE_CONFIG_JSON",
      streamGroup: "performance",
    },
  ],
});

Use one destinationSchema for each enabled Sync Connection. This isolates Airbyte Direct Load tables and generation state across accounts and catalogs.

resolve the secret at the deployment boundary

Load the named secret in your Worker or operator process. Pass its parsed provider configuration to syncConnectionDefinitionFromConfig(). The package uses the value to derive non-secret identity and fingerprints; the desired-state config keeps only the secret name.

const connection = await syncConnectionDefinitionFromConfig(
  config.connections[0],
  parsedSourceSecret
);

Do not log or commit parsedSourceSecret.

validate the connection set

validateDestinationSchemaIsolation(connections);

The validator throws when two enabled connections claim the same Airbyte schema. Treat that as a configuration error and choose distinct schemas before a run.

run the complete tutorial

The public source tutorial defines Google Ads, Meta performance, and Meta metadata connections, resolves placeholder source identities, and validates schema isolation:

pnpm --filter @patronage/ads-sync build
node --experimental-strip-types packages/ads-sync/src/tutorials/configure-connections.ts

Read and run configure-connections.ts. Continue with verify a bounded run before adapting a reference deployment to real data.