Configure the Model in agent

[xyzhu68]

Hi,
in an Agent you can define an LLM, e.g. deepseek-chat. You can also define the system prompt (I think this is for the LLM or is it for the agent or both?)
My question is, is it possible to make function calling with the LLM? How can I define "tools" in the request and how can I get "tool_calls" from the response of the LLM?

[Llala0928]

@xyzhu68 1. System prompts are for LLM, and the Agent framework is only responsible for forwarding. 2. Prompts are needed to tell the Agent how to use the tool.