Prompts are defined via a custom post type (similar to Create a Custom Post Type in Hub). Therefore, you can find them near the top of the left-side menu in the Purple Hub and can manage them similarly to how posts are managed.
Clicking on 'Purple Prompts' (1) > 'All Prompts' (2) shows the list of all prompts defined in your Purple Hub:
When editing a prompt, you find no block editor, but instead a simple text box next to some additional settings.
The text box (1) is used to define the prompt, and you can use placeholders to add e.g. the post content into the prompt at the fitting location.
On the right side, you need to define firstly the intended usage area for this prompt (2). The prompt can then be used for this feature only. As an example, this prompt here asks for a specific tone and style of the post title, and therefore it should only be used for the title generation feature (see Generating Meta Information) and not e.g. article rewriting.
Below that are settings that affect the language model itself:
- Max Tokens (4): This is a parameter that controls the maximum length of the generated text. Text generation will automatically stop at the given limit.
'Publishing' (1) a prompt makes it available for editors. If a prompt is unpublished, in draft mode, or gets trashed, it can not be used for its configured use case and will not be shown there. This can be helpful when you want to experiment with many different prompts and clean up afterwards without permanently deleting the prompts.