Staging noise shouldn't page production on-call, and a developer's API key shouldn't be able to touch production monitors. Environments tag every monitor, credential and API key, and API keys can be restricted to the environments they may act on.
Create the environments your organization actually has — production, staging, dev, per-customer — and assign monitors to them. Alert destinations and incident rules can then differ per environment, so staging posts to a channel while production pages.
On the access side, an API key or service account can be limited to specific environments; a CI pipeline that creates staging monitors physically cannot create or delete production ones.