Module provider 

Provider trait and types for LLM abstraction.

This module defines the LlmProvider trait that all LLM providers implement, along with request/response types, tool definitions, and configuration.

Overview

The provider abstraction allows you to:

• Switch between LLM providers without changing your application code
• Use a consistent interface for all LLM operations
• Access provider-specific features (like caching) through unified APIs

Provider Trait

All providers implement LlmProvider, which provides:

• execute_llm(): Execute a standard LLM request
• execute_structured_llm(): Execute with JSON schema output
• provider_name(): Get the provider identifier

Tool Calling

Define tools with Tool and handle the calling flow:

use multi_llm::{Tool, ToolChoice, ToolCall, ToolResult};

// Define a tool
let weather_tool = Tool {
    name: "get_weather".to_string(),
    description: "Get current weather for a city".to_string(),
    parameters: serde_json::json!({
        "type": "object",
        "properties": {
            "city": {"type": "string", "description": "City name"}
        },
        "required": ["city"]
    }),
};

// Handle a tool call from the LLM
let tool_call = ToolCall {
    id: "call_123".to_string(),
    name: "get_weather".to_string(),
    arguments: serde_json::json!({"city": "London"}),
};

// Return the result
let result = ToolResult {
    tool_call_id: "call_123".to_string(),
    content: "Sunny, 22°C".to_string(),
    is_error: false,
    error_category: None,
};

Response Structure

All providers return a Response containing:

• Text content (for standard requests)
• Structured JSON (when using execute_structured_llm)
• Tool calls (when the model wants to call functions)
• Token usage statistics

Structs

RequestConfig

Configuration for a single LLM request.

Response

Response from an LLM operation.

ResponseFormat

Schema specification for structured JSON output.

TokenUsage

Token usage statistics for an LLM request.

Tool

Definition of a tool/function that the LLM can call.

ToolCall

A tool call generated by the LLM.

ToolCallingRound

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

ToolResult

Result from executing a tool, sent back to the LLM.

Enums

ToolChoice

Strategy for how the LLM should handle tool selection.

Traits

LlmProvider

Trait implemented by all LLM providers.

Type Aliases

LLMRequestConfig

Type aliases for backward compatibility

LLMResponseFormat

LLMTokenUsage

Result

Result type alias for provider operations.