Struct ToolCallingRound 

pub struct ToolCallingRound {
    pub assistant_message: UnifiedMessage,
    pub tool_results: Vec<ToolResult>,
}

State for a tool calling round in multi-turn conversations.

When using tool calling, conversations often have multiple rounds:

1. User asks a question
2. Assistant requests tool calls
3. Tools execute and return results
4. Assistant uses results to form final response

ToolCallingRound captures the assistant’s tool requests and the corresponding results, allowing providers to properly format multi-turn tool conversations.

Example Flow

use multi_llm::{ToolCallingRound, ToolResult, UnifiedMessage};

// After receiving tool calls from the LLM
let round = ToolCallingRound {
    assistant_message: assistant_response,  // The message with tool calls
    tool_results,  // Results from executing those calls
};

// Pass to execute_llm for the next turn
// provider.execute_llm(request, Some(round), config).await?;

Fields

assistant_message: UnifiedMessage

The assistant message that initiated tool calls.

This message contains the tool call content variants that the assistant generated.

tool_results: Vec<ToolResult>

Results from executing the requested tools.

Each result’s tool_call_id should match an ID from the assistant message.

Trait Implementations

impl Clone for ToolCallingRound

fn clone(&self) -> ToolCallingRound

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ToolCallingRound

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.