Specifically, in the Decimal reference definition:

http://build.fhir.org/ig/HL7/cql/09-b-cqlreference.html#decimal-1

Add an implementers note (yellow box):

Note that implementations must support at least the 28 digits of precision and 8 digits of scale, but may support more precise values.

http://build.fhir.org/ig/HL7/cql/09-b-cqlreference.html#maximum

Add an implementers note (yellow box, after the type-specific descriptions, before the examples):

Note that if implementations support larger and/or more precise values than the minimum required precision and scale for Decimal, DateTime, and Time values, they will return the maximum representable decimal for the implementation.

http://build.fhir.org/ig/HL7/cql/09-b-cqlreference.html#minimum

Add an implementers note (yellow box, after the type-specific decriptions, before the examples):

Note that if implementations support larger and/or more precise values than the minimum required precision and scale for Decimal, DateTime, and Time values, they will return the minimum representable decimal for the implementation.

http://build.fhir.org/ig/HL7/cql/09-b-cqlreference.html#successor

Add an implementers note (yellow box, after the type-specific descriptions, before the examples):

Note that implementations that support more precise values than the minimum required precision and scale for Decimal, DateTime, and Time values, the successor will reflect the minimum representable step size for the implementation.

http://build.fhir.org/ig/HL7/cql/09-b-cqlreference.html#predecessor

Add an implementers note (yellow box, after the type-specific descriptions, before the examples):

Note that implementations that support more precise values than the minimum required precision and scale for Decimal, DateTime, and Time values, the predecessor will reflect the minimum representable step size for the implementation.

http://build.fhir.org/ig/HL7/cql/09-b-cqlreference.html#todecimal

Replace the paragraph beginning with "Note that the decimal value..." with:

Note that the decimal value returned by this operator will be limited in precision and scale to the maximum precision and scale representable by the implementation (at least 28 digits of precision, and 8 digits of scale).

Reflect these same clarifications in the Logical operator definitions (which means in the ELM schema definitions, as well as in the Logical Specification (Chatper 4) documentation of the ELM nodes for these operators.