A settings panel for configuring AI agent parameters including temperature, max tokens, top-p, penalties, system prompt, stop sequences, and streaming toggle. Available in full (AgentSettings) and simplified (QuickAgentSettings) variants. Supports collapsible sections and custom labels.
Controls randomness. Lower is more deterministic.
Maximum number of tokens to generate
Alternative to temperature, nucleus sampling
Instructions that guide the model's behavior
Stream responses as they generate