Question about Tool Call Information in Multi-Turn Conversations

[glimmera]

Description

I'm working with multi-turn conversations in Spring AI and have a question regarding how tool call information is handled and persisted within the conversation history.

Specifically, I'd like to clarify two points:

1. Tool Call Storage in Conversation History: When a tool is invoked during the first turn of a multi-turn conversation, is the information about this tool call (e.g., the tool's name, the arguments passed to it, and its output) stored as part of the overall conversation history?
2. Tool Call Information in Subsequent Turns: If the tool call information is stored in the history, is this information automatically included when making a request for the second or subsequent turns of the conversation? In other words, does the model receive context about previous tool invocations when continuing the dialogue?

---

Additional Question:

Is there a programmatic way within Spring AI to inspect or retrieve this tool call information from the conversation history at any point during the multi-turn interaction? This would be helpful for debugging or for implementing custom logic based on past tool usage.

---

Thank you for your time and clarification!