v0.3.0

Changed

• LLM temperature is now opt-in — the node's Temperature control defaults to Auto, leaving the value unset so the model uses its own default. This means models that reject a temperature parameter work out of the box. Switch to Custom to set a specific value.

Added

• acceptsTemperature option on createOpenAIProvider — set it false to omit temperature for a model family that rejects it.
• A clear, actionable error when a model rejects a temperature you've set, instead of a raw provider 400.