Prompt chaining is an advanced technique enabling sequential execution of multiple prompts to combine several story tools into one comprehensive tool. This capability allows writers to streamline complex workflows by chaining up to three steps together.
Setting Up a Prompt Chain
To create a prompt chain, open the story tool creator and locate the “+Add Step” button in the bottom left corner of the interface. Clicking this button adds another prompt box where you can input instructions for each step. Tools support up to three sequential steps.
Key Features
Each step functions independently and can:
- Use different AI models (e.g., an OpenAI model for Step 1, an Anthropic model for Step 2)
- Reference the same or different context sources
- Process outputs from previous steps
Critical requirement: For Steps 2 and 3 to work, the output from the prior step needs to be included in the later step’s prompt context. This prior output appears as an available reference option in the “Add reference” dropdown menu.
Example: Two-Step Revision Workflow
A practical implementation involves:
Step 1 — Use an OpenAI model to analyze a chapter for plot holes and logical inconsistencies, providing detailed feedback and reasoning.
Step 2 — Use an Anthropic model to rewrite the chapter implementing the identified changes while preserving unmodified content.
Executing a Prompt Chain
During execution, multi-step tools run sequentially. The output box displays dots corresponding to the number of steps (two dots for two-step tools, three for three-step tools). After completion, clicking these dots reveals each stage’s output.
Important: Only one output can be inserted into the text, so review each step’s results carefully before selecting “Keep it.”