​

Getting Tools

The xpander SDK allows you to retrieve tools connected to your agent in the xpander platform.

​

Request

​

Response

The tools are returned in the format required by your LLM provider. Here’s an example response for different providers:

​

Run Tools

To use tools with your LLM, follow these steps:

1. Get the tools from your agent
2. Pass them to your LLM
3. Extract and execute any tool calls
4. Process the results

llm_response = openai_client.chat.completions.create(
    model="gpt-4",
    messages=conversation_history,
    tools=tools
) 

​

Function Calling Methods

There are two main approaches to executing tools. The SDK supports both parallel and sequential execution of tools

1. Automated LLM Parsing (Recommended):

# Parse LLM response automatically into tool calls
tools_to_run = XpanderClient.extract_tool_calls(llm_response=llm_response.model_dump())

# Execute multiple tool calls at once
results = agent.run_tools(tools_to_run)  

2. Manual Invocation:

from xpander_sdk import ToolCall, ToolCallType

# Create a ToolCall object
toolcall = ToolCall(
    name="tavily-insights-fetchInsightsFromTavilyAI",
    type=ToolCallType.XPANDER,
    payload={
        "bodyParams": {
            "query": "qwen2.5-coder news on HackerNews",
            "max_results": "5",
            "include_raw_content": "true",
            "include_images": "false",
            "exclude_domains": ["example.com"], 
        },
        "queryParams": {},
        "pathParams": {}
    }
)

# Execute single tool call
result = agent.run_tool(toolcall)

​

Tool Classes

The SDK provides two main classes for working with tools:

​

ToolCall

A class representing a single tool invocation with the following attributes:

• name: (str) The identifier of the tool to be called
• type: (ToolCallType) The type of the tool call, usually ToolCallType.XPANDER
• payload: (dict) Contains three parameter dictionaries:
  • bodyParams: Parameters sent in the request body
  • queryParams: URL query parameters
  • pathParams: URL path parameters

​

ToolCallType

An enum class that defines the types of tool calls:

• XPANDER: Standard xpander.ai platform tools
• FUNCTION: Traditional function calls (For example local functions)

​

Tool Response

The tool_response object represents the output from a tool invoked by an AI Agent. It contains key attributes that provide information about the tool invocation’s execution and results. This data is used to build the AI Agent’s memory and guide further interactions.

​

Tool Response Object Structure

​

Available Providers and Models

Here are all the supported LLM providers and their corresponding models. Choose the combination that best fits your needs: