Prompt Helper

The Prompt Helper is an AI-powered node that writes and improves prompts for you. Connect it between a source node and a generation node, and it reads the available context — images, text, files — to produce a prompt that's ready to use downstream.

Two modes and how to use it

Compose — the Prompt Helper reads all connected upstream context and writes a prompt from scratch. Use this when you have a rough concept or an image you want to build on, but aren’t sure how to phrase the prompt.

Enhance — you write a draft prompt and the Prompt Helper rewrites it to be more effective: more specific, more visually descriptive, and better calibrated for the selected downstream model.

How to use it:

  1. Add a Prompt Helper node to the canvas

  2. Choose a mode: Compose or Enhance

  3. Optionally connect upstream nodes (Image, Text, File, Video, or Audio) as context inputs

  4. Optionally set a System Prompt to give the helper a specific instruction or persona

  5. Select which LLM model should power the helper

  6. Click Generate — the output is a text prompt you can wire into any generation node

Context inputs and example workflow

The Prompt Helper automatically reads outputs from any connected upstream nodes:

  1. Image node — visual description and generation metadata

  2. Text node — the full text output

  3. File node — uploaded document or image content

  4. Audio node — transcription or audio description

  5. Video node — frame descriptions

Example workflow: Connect a File node (product photo) to the Prompt Helper set to Compose mode with system prompt “Write a clean, minimal product photography prompt”. Wire the output into an Image node (Flux 2 Max). The Prompt Helper reads the uploaded photo, composes a matching prompt, and the Image node generates a new visual in the same style.