> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rubbrband.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompting Nodes

> Learn about the various Prompting Nodes that Rubbrband has

## Text Generator Node

The text generator node produces text given a prompt and a task. It's good for workflows that require an input prompt to be modified in a certain way every time. For instance, if you want to make all input prompts into "pirate-style" prompts.

**Inputs**

<ResponseField name="prompt" type="string" required={false}>
  The input text that is used as context to generate the output text. If not passed in, the node will only consider the task definition in the settings.
</ResponseField>

**Settings**

<ResponseField name="task" type="string">
  A description of the task -- this describes what the GPT is supposed to do.&#x20;

  Example: "You are a pirate image generator, create a prompt inspired from pirates for a stable diffusion text to image, given a topic that I will provide to you. Please make sure the prompt is clear and concise."

  In this case, the topic would come from the prompt.
</ResponseField>

**Outputs**

<ResponseField name="Text" type="string">
  The text response from the language model
</ResponseField>

## Describe Node

The describe node produces text given an image input. It can be used to describe the image, but also for any task that requires viewing the image to produce text.

**Inputs**

<ResponseField name="Image" type="image" required={true}>
  The image that is provided as context for the task
</ResponseField>

**Settings**

<ResponseField name="Prompt" type="string">
  The proposed prompt for the language model to respond to. For example: "Describe this image, but imagine that the background is a medieval town, instead of what it is right now. Provide the description for the image with the new background."
</ResponseField>

**Outputs**

<ResponseField name="Text" type="string">
  The text response from the language model
</ResponseField>