An AI practitioner is creating a reusable prompt in Amazon Bedrock Prompt Management that must call external APIs or services during execution. Which feature should they use to enable calling external services?

Defining a tools configuration is the correct approach because Amazon Bedrock Prompt Management uses tools to integrate with external APIs or services. This allows the prompt to invoke specific functions or retrieve information from outside sources during its execution. Inserting special tokens into the prompt might be used for specific model instructions but doesn't inherently enable external API calls. Using prompt variables passes dynamic data into the prompt, but doesn't facilitate calling external services. A stop sequence controls when the model stops generating output, which is unrelated to external service integration.