Agent node

Agent Name

The agent's name is primarily used to distinguish between agents. It also serves as a prefix when the agent's output is stored as variables.

When the agent name is converted to a variable prefix, it is transformed to snake_case. For example, if the agent's name is Test Agent, it becomes test_agent.

Select Model

Select a model to be executed in the agent. All available models, based on the connected providers, are listed for you to choose.

Agent Panels

You can configure the agent's details in the following panels:

📄️ Prompt

A prompt is the core input value passed to the model. In the context of Large Language Models (LLMs), a prompt is a text instruction or question that guides the model to generate a desired response. It serves as the primary way to communicate with the AI model and influence its behavior and output.

📄️ Parameters

Parameters are configuration settings that control how the LLM generates responses. These settings influence the model's behavior, creativity, randomness, and output characteristics. By adjusting parameters, you can fine-tune the model's responses to match your specific needs, whether you want more creative and varied outputs or more focused and deterministic results.

📄️ Output

You can determine the format in which the agent responds and whether to use streaming.

📄️ Preview

The Preview panel allows you to see exactly how the prompt and parameters will appear when sent to the model. This is essential for testing and debugging your agent configuration before running it in a live session. You can verify that variables are being populated correctly, messages are formatted properly, and parameters are set as expected.