We have an account-level pipeline template used by many repos.

That pipeline template is composed of multiple step group templates (for example: curate inputs, initialize pipeline, centralized selector, pants/native test/build groups, GraphQL groups).

Consumer repos define a linked pipeline.yaml and pass templateInputs.variables only for values they explicitly want to override.

Our goal is:

Keep consumer pipeline.yaml files lean (only intentional overrides).

Let template defaults apply when variables are omitted.

Still allow per-pipeline overrides later by adding a variable in the consumer pipeline.

Avoid reconciliation repeatedly adding/injecting many defaulted fields into consumer YAML.

What we observed in testing:

If template variable is fixed literal:

Omit from consumer: works (template default applies).

Add in consumer as override: not allowed / reconciled away (expected, since not an input contract).

If template variable is set with <+input>.default(...).executionInput():

Override works.

But reconciliation materializes those runtime-input fields in linked pipelines, which makes consumer YAML less lean.

We are looking for a better way to override these template variables without having to constantly reconcile the entities calling the templates.