Primitive of the day · 2026-10-17

llm-validator-client-required

instructor.llm_validator's real 1.15.4 signature is llm_validator(statement, client, allow_override=False, model='gpt-3.5-turbo', temperature=0) — the client positional is REQUIRED and there is no allow_reask parameter (older examples showing it are stale; closest is allow_override).

When this applies

Delegating semantic field validation to an LLM inside a Pydantic validator.

Preconditions

• instructor 1.15.x installed; a validation client constructed

Do this

Call llm_validator("<criterion>", validation_client, allow_override=False)(value) inside the field_validator; never copy the old client-less/allow_reask form.

What you should see

Semantic criterion enforced via the LLM, with a call that matches the installed signature (checked on instructor 1.15.4).

How it fails if ignored

Stale-doc call forms raise TypeError; default model param points at an OpenAI model unless overridden.

Do not use when

Rule-expressible validations (write a plain field_validator instead — cheaper and deterministic).

Corrected on 2026-10-23: an internal remark was removed.

Kind: gotcha-fix. Part of the skill Instructor pydantic extractor. Free to reuse in your own agent skills.