A production data pipeline should not depend on a departed employee's personal token. Which identity approach should be evaluated?

1. A table description used as a credential
2. Anonymous write access
3. A governed service principal or appropriate supported workload identity
4. One shared personal token copied to all repositories