lb.

Prompting & Structured Output

Prompts as API contracts: system prompt design, few-shot patterns, JSON schema outputs, tool definitions, and treating model output as untrusted input.

  1. 01·The System Prompt Is an API ContractThe system prompt is your service's interface definition. Structure it like config, separate instructions from data, and set precedence rules.
  2. 02·Examples Are Fixtures You Ship on Every RequestFew-shot examples are test fixtures you ship on every request. When examples beat instructions, how to pick them, and what they cost in tokens.
  3. 03·Prompts Are Config That Can Take Down ProdPrompts are hot-path config written in English. Version them in git, deploy them like code, log the version on every call, and roll back in one step.
  4. 04·Three Ways to Get JSON Out of a ModelJSON mode, structured outputs, and the forced tool call solve different problems. What each guarantees, what it doesn't, and which one to reach for.
  5. 05·Schemas the Model Can Fill, Tools the Model Can CallEnums over free strings, flat over nested, descriptions as docs. Design schemas a model can fill, then define your first tool the same way.
  6. 06·Model Output Is Untrusted InputSchema-valid is not correct. Validate model output like user input with Zod, repair failures once, then compose it all into a typed LLM function.