Function Schema

A function schema is the machine-readable description of a tool: its name, what it does, and a JSON Schema for its parameters, including which are required and what values are allowed. It is what a model reads before deciding whether and how to call something. It does two jobs at once, which is why it is worth writing carefully. It is documentation aimed at a reader that cannot ask clarifying questions, and it is a validation contract your own code can enforce before executing anything. Enums beat free-text strings, required fields beat optional ones with defaults, and a parameter that can only sensibly take three values should say so in the schema rather than in the description. The common failure is a schema that is technically valid and semantically vague. "query: string" gives the model no idea what a good query looks like; "query: string — a full-text search over invoice line items, e.g. 'unpaid invoices from March'" produces calls you can actually serve. Version schemas the way you version an API. A field that changes meaning silently breaks every prompt that learned the old meaning.