A record can pass a structural check and still contain an inaccurate observation. A schema helps define what a record must look like; it does not independently verify the event the record describes.
For a fictional reading log, suppose a record needs a text title and an integer page count. A schema could reject a list placed where the title should be. It cannot, from that structure alone, establish that somebody read the book or copied its page count correctly.
Presence and type are separate questions
In JSON Schema, defining a property’s rules is not the same as requiring the property to appear. The official introduction shows how property definitions and required fields work together.
Make a tiny set of examples: one valid record, one missing a required field, and one containing the wrong type. Check that the validator responds as you expect. Keep the schema version with the examples.
Then list checks that still need another process: source verification, plausible ranges, duplicate handling or review of unusual entries. Some can be encoded as additional rules; others require evidence outside the record.
Passing validation should mean “passed these stated rules.” Giving it that precise meaning makes the check useful without asking it to promise more than it can deliver.