HeroML is an AI Prompt Chain/Workflow interpreter for Apps built on https://hero.page
HeroML v0.0.1 Documentation
HeroML (Hero Markup Language) is a novel approach for setting up multi-step workflows to interact with AI models like OpenAI's GPT-3 and GPT-4.
In this version, you can set up workflows in a series of steps, where each step is a prompt that the AI will respond to. The responses from previous steps can be used in the prompts for later steps.
Each workflow is a series of steps separated by the "->>>>" delimiter.
In the first step, you can define dynamic variables in the format
Using Responses from Previous Steps
You can reference the response of a previous step in your current step's prompt using the format
There's only one action available in v0.0.1, which is "Loop". This action can be used to iterate over each item in an array from the response of a previous step.
In the first line of the prompt, write
In the example above,
Format of the Response
You must specify the format you expect the response to be in. For example, if you want the response to be an array of strings, you can specify "Return the list as an array of strings".
Here's an example of a workflow:
In this workflow, the first step is a prompt asking the AI to write a blog post about a certain topic, including three points. The second step is a loop action that asks the AI to write a paragraph about each point. The result of the second step will be an array of paragraphs, each about one point from the first step.
This documentation provides a basic overview of how to create and use workflows in HeroML v0.0.1. As the language evolves, new features and functionalities will be added. For the most up-to-date information, always refer to the latest documentation.