Completing the Tool Use Cycle with GPT-5

Introduction & Overview

In the previous lessons, you learned how to create tool schemas and understand GPT-5's responses when it wants to use tools. You can now recognize when GPT-5 requests tool execution by checking for type: "function_call" items in the output array and extracting the necessary details from function call items. However, knowing what GPT-5 wants to do is only half the story — you still need to actually execute those tools and complete the conversation cycle. In this lesson, you'll learn how to bridge that gap by executing the functions GPT-5 requests, capturing their results, and sending those results back to GPT-5 in the proper format. By the end of this lesson, you'll have a complete tool execution pipeline that can handle GPT-5's tool requests from start to finish, maintaining proper conversation flow throughout the entire process.

The Complete Tool Execution Flow

Before we dive into the implementation, let's understand the complete workflow we'll be building in this lesson. Here's the step-by-step process that transforms GPT-5 from a simple chatbot into a capable agent: Set up the foundation - Create function mappings, tool schemas, and initial conversation messages Send the initial request - Make the first API call to GPT-5 using responses.create with the user's question and available tools Detect tool use requests - Filter through the output array to find items with type: "function_call" Extract tool information - Pull out the function name, arguments, and call ID from each function call item Execute the requested functions - Parse the JSON arguments and use our function mapping to call the actual TypeScript functions with GPT-5's parameters Collect and format tool results - Structure all function outputs in the function_call_output format with matching call_id values Send results back to GPT-5 - Make a second API call with the complete conversation, including both the original function_call items AND their outputs Display the final response - Show GPT-5's natural language answer that incorporates the tool outputs This complete cycle enables GPT-5 to seamlessly use tools as part of its reasoning process, transforming raw function outputs into conversational responses that directly answer user questions.

Setting Up the Foundation

As covered in previous lessons, we need to map tool names to functions and prepare our tool schemas. Here's the complete setup that enables tool execution: TypeScriptimport fs from "fs"; import OpenAI from "openai"; import { sumNumbers, multiplyNumbers } from "./functions"; // Initialize the OpenAI client const client = new OpenAI(); // Create an object mapping tool names to functions // The keys must match our tool schema names exactly const tools: Record<string, Function> = { sum_numbers: sumNumbers, multiply_numbers: multiplyNumbers }; // Choose a model to use const model = "gpt-5"; // System instructions for the assistant const systemPrompt = "You are a helpful math assistant. " + "Always use the available tools to perform calculations accurately."; // Load the schemas from JSON file const schemasJson = fs.readFileSync("schemas.json", "utf-8"); const toolSchemas = JSON.parse(schemasJson); // Create an array of messages to send to GPT-5 const messages: Array<Record<string, unknown>> = [ { role: "user", content: "Please calculate 15 + 27" } ];import fs from "fs"; import OpenAI from "openai"; import { sumNumbers, multiplyNumbers } from "./functions"; // Initialize the OpenAI client const client = new OpenAI(); // Create an object mapping tool names to functions // The keys must match our tool schema names exactly const tools: Record<string, Function> = { sum_numbers: sumNumbers, multiply_numbers: multiplyNumbers }; // Choose a model to use const model = "gpt-5"; // System instructions for the assistant const systemPrompt = "You are a helpful math assistant. " + "Always use the available tools to perform calculations accurately."; // Load the schemas from JSON file const schemasJson = fs.readFileSync("schemas.json", "utf-8"); const toolSchemas = JSON.parse(schemasJson); // Create an array of messages to send to GPT-5 const messages: Array<Record<string, unknown>> = [ { role: "user", content: "Please calculate 15 + 27" } ]; The tools object is the key component here — it enables dynamic function lookup when GPT-5 requests tool execution by name.

Join the 1M+ learners on CodeSignal

Be a part of our community of 1M+ users who develop and demonstrate their skills on CodeSignal

Detecting Tool Use Requests

After receiving GPT-5's response, we need to examine the output array to determine if tool execution is needed. Unlike checking a single stop_reason field, we filter through the output items and look for function call requests.

TypeScript

// STEP 2: Look for function calls in the response
// Function calls appear as items with type "function_call" in the output
const functionCalls = response.output.filter(
  (item) => item.type === "function_call"
);

// STEP 2: Look for function calls in the response
// Function calls appear as items with type "function_call" in the output
const functionCalls = response.output.filter(
  (item) => item.type === "function_call"
);

The response.output array contains all output items from GPT-5, which can include text messages, function calls, and other content types. By using the .filter() method and checking each item's type property for "function_call", we identify all the tools GPT-5 wants to execute. The filter returns a new array containing only the function call items, which might be empty if GPT-5 doesn't need any tools, or contain multiple items if GPT-5 requests several function calls.

Once we've collected all function calls, we check if any were found and proceed accordingly:

TypeScript

// Check if GPT-5 wants to call any functions
if (functionCalls.length > 0) {
  // STEP 3: Execute each function call that the model requested
  console.log("Executing function calls...");
  
  // Initialize arrays to store function call items and outputs
  const functionCallItems: Array<Record<string, unknown>> = [];
  const functionOutputs: Array<Record<string, unknown>> = [];
  
  for (const functionCall of functionCalls) {
    if (functionCall.type !== "function_call") continue;
    
    // Get the function name, arguments, and call ID
    const toolName = functionCall.name;
    const callId = functionCall.call_id;
    
    // Parse arguments from JSON string to object
    const toolInput = JSON.parse(functionCall.arguments);

// Check if GPT-5 wants to call any functions
if (functionCalls.length > 0) {
  // STEP 3: Execute each function call that the model requested
  console.log("Executing function calls...");
  
  // Initialize arrays to store function call items and outputs
  const functionCallItems: Array<Record<string, unknown>> = [];
  const functionOutputs: Array<Record<string, unknown>> = [];
  
  for (const functionCall of functionCalls) {
    if (functionCall.type !== "function_call") continue;
    
    // Get the function name, arguments, and call ID
    const toolName = functionCall.name;
    const callId = functionCall.call_id;
    
    // Parse arguments from JSON string to object
    const toolInput = JSON.parse(functionCall.arguments);

When we find function call items, we extract the essential information from each one: the function name (which function to call), the call_id (a unique identifier we'll need to match results back to requests), and the arguments (provided as a JSON string). The JSON.parse() call is essential because the arguments come as a JSON string, not a TypeScript object, so we must parse them before we can use them to call our functions.

Executing the Requested Functions

With the function information extracted and arguments parsed, we can now execute the actual TypeScript functions using our mapping object. This is where the tool names from GPT-5's requests get translated into actual function calls.

TypeScript

if (functionCalls.length > 0) {
  // We have function calls to execute
  console.log("Executing function calls...");
  
  // Initialize arrays to store function call items and outputs
  const functionCallItems: Array<Record<string, unknown>> = [];
  const functionOutputs: Array<Record<string, unknown>> = [];
  
  for (const functionCall of functionCalls) {
    if (functionCall.type !== "function_call") continue;
    
    // Extract the function name, arguments, and call ID
    const toolName = functionCall.name;
    const callId = functionCall.call_id;
    
    // Parse arguments from JSON string to object
    const toolInput = JSON.parse(functionCall.arguments);
    
    // Print the tool name and input
    console.log(`Executing: ${toolName}(${JSON.stringify(toolInput)})`);
    
    // Use try-catch to handle both missing tools and execution errors
    let result: any;
    try {
      // Check if the requested tool exists and execute it
      if (toolName in tools) {
        result = tools[toolName](toolInput);
      } else {
        result = `Error: Function ${toolName} not found`;
      }
    } catch (e: any) {
      // Handle any errors that occur during function execution
      result = `Error executing ${toolName}: ${e.message}`;
    }
    
    // Print the result
    console.log(`Result: ${result}`);

The critical step here is parsing the arguments with JSON.parse(functionCall.arguments) before passing them to the function. Without this parsing step, we'd be trying to use a JSON string instead of an object, which would cause an error. Using a try-catch block provides comprehensive error handling that catches not only missing tools but also any runtime errors that might occur during function execution. For example, if a function receives invalid parameters (like trying to divide by zero), the exception will be caught and converted into an error message that can be sent back to GPT-5. This feedback mechanism allows the model to reason about the failure and decide on a next step, such as retrying with corrected parameters, using an alternative tool, or explaining the issue to the user, rather than simply failing.

Formatting and Adding Tool Results to the Conversation

After executing each tool, we must format both the function call and its result, then add them to the messages array. The Responses API requires both pieces in specific formats.

TypeScript

    // Add the function call item
    functionCallItems.push({
      type: "function_call",           // Identifies this as a function call
      name: toolName,                  // The name of the function that was called
      arguments: functionCall.arguments,  // The original arguments as JSON string
      call_id: callId                  // Unique identifier for this function call
    });
    
    // Package the result in the format the Responses API expects
    functionOutputs.push({
      type: "function_call_output",  // Special content type for function results
      call_id: callId,               // Links result to original request
      output: JSON.stringify({ result })  // The actual function output as JSON
    });
  }
  
  // STEP 4: Add the function calls and their results to our conversation
  messages.push(...functionCallItems);
  messages.push(...functionOutputs);

    // Add the function call item
    functionCallItems.push({
      type: "function_call",           // Identifies this as a function call
      name: toolName,                  // The name of the function that was called
      arguments: functionCall.arguments,  // The original arguments as JSON string
      call_id: callId                  // Unique identifier for this function call
    });
    
    // Package the result in the format the Responses API expects
    functionOutputs.push({
      type: "function_call_output",  // Special content type for function results
      call_id: callId,               // Links result to original request
      output: JSON.stringify({ result })  // The actual function output as JSON
    });
  }
  
  // STEP 4: Add the function calls and their results to our conversation
  messages.push(...functionCallItems);
  messages.push(...functionOutputs);

We reconstruct each function call item because the original functionCall objects are SDK model objects, not plain JSON objects. The Responses API requires a strictly shaped payload with only the allowed fields (type, name, arguments, call_id).

For function outputs, the call_id must exactly match the original function call's ID, and the output must be a JSON string. It's critical to provide an output for every function call — if you skip any call_id, your next API call will fail because OpenAI expects results for all requested functions.

Finally, we add both the function calls and their outputs to the messages array using the spread operator (...), maintaining the proper conversation structure that GPT-5 expects.

Getting GPT-5's Final Response

Once we've executed all requested functions and added both the function calls and their results to the messages array, we can send the updated conversation back to GPT-5 to get the final response that incorporates the tool results.

TypeScript

  // STEP 5: Send everything back to the model to get the final answer
  // Now the model can use the function results to formulate its response
  const finalResponse = await client.responses.create({
    model,
    instructions: systemPrompt,
    input: messages,  // Now includes both function calls and their outputs
    tools: toolSchemas,
    reasoning: { effort: "minimal" },
    store: false
  });
  
  // Print the model's final answer (which should use our function results)
  console.log("\nGPT-5's final response:");
  console.log(finalResponse.output_text);

  // STEP 5: Send everything back to the model to get the final answer
  // Now the model can use the function results to formulate its response
  const finalResponse = await client.responses.create({
    model,
    instructions: systemPrompt,
    input: messages,  // Now includes both function calls and their outputs
    tools: toolSchemas,
    reasoning: { effort: "minimal" },
    store: false
  });
  
  // Print the model's final answer (which should use our function results)
  console.log("\nGPT-5's final response:");
  console.log(finalResponse.output_text);

This second API call uses the same parameters as the first, but now the input array contains the complete conversation history, including both the original function call items and their corresponding outputs. GPT-5 can now provide a comprehensive answer that incorporates the tool execution results, transforming raw calculation outputs into natural, conversational responses.

The output_text property provides a convenient way to extract just the text content from GPT-5's response, filtering out any non-text items in the output array.

The complete execution flow produces the following output, showing both the tool execution and GPT-5's final response:

text

Executing function calls...
Executing: sum_numbers({"a":15,"b":27})
Result: 42

GPT-5's final response:
15 + 27 = 42

Executing function calls...
Executing: sum_numbers({"a":15,"b":27})
Result: 42

GPT-5's final response:
15 + 27 = 42

This output demonstrates the complete tool execution cycle: our code executes the requested function with the provided parameters, captures the numerical result, and then GPT-5 transforms that raw output into a natural, conversational response that directly answers the user's original question.

Introduction & Overview

The Complete Tool Execution Flow

Setting Up the Foundation

Join the 1M+ learners on CodeSignal

Sending the Initial Request

Detecting Tool Use Requests

Executing the Requested Functions

Formatting and Adding Tool Results to the Conversation

Getting GPT-5's Final Response

Handling Non-Tool Responses

Visualizing the Complete Conversation History

Summary & Practice Preparation