![Controlling the content of text sections in the [gpt_article] shortcode](https://www.cyberseo.net/wp-content/uploads/2024/05/controlling-the-content-of-text-sections.png "Controlling the content of text sections in the [gpt_article] shortcode")
The [gpt_article] shortcode used in the CyberSEO Pro and RSS Retriever plugins has introduced a new parameter that allows users to provide additional directives for the prompts used to generate each individual article section, similar to the functionality offered by the AI Autoblogger plugin.
Unlike AI Autoblogger, where the prompts are entered as a plain text document, CyberSEO Pro and RSS Retriever work with shortcodes. This requires a different way of formatting the directives used to generate each section of text, but achieves the same result. Let’s take a closer look at this new parameter, which has the following syntax: sectionN="your section-specific directives".
Where N is the numeric section number of the article section for which you want to specify your own rules, in addition to the main prompt set by the directives parameter. This approach allows for a high degree of customization and lets you tailor content generation to specific sections of the article, improving relevance and consistency based on the unique needs of each section.
In your shortcode configuration, you can either choose not to include any sectionN parameters, or to include one or more of them. In the first case, the same prompt specified by your main directives parameter will be used to generate all text sections of the article. In the second case, the values of directives and sectionN are combined into a single unified prompt for each specific section. For example:
[gpt_article engine="anthropic/claude-3-opus" sections="4" topic="%post_title%" directives="The article should tell about known facts reviving the secrets." section2="Format the text for embedding in a WordPress post, using only <p>, <em>, <blockquote>, <ul>, <li> tags. Add one quote from your knowledge base. Exclude any other HTML tags." section4="Format the text for embedding in a WordPress post, using only <p>, <em>, <table>, <tr>, <td> tags. Include one table with a useful information. Exclude any other HTML tags. Tell about a mystery or secret related to the topic and invite the reader to find the answer for themselves. Encourage them to think."]
In this setup, the prompts for generating the different sections of the article are intended to provide specific instructions that improve the relevance and formatting of each section, ensuring a cohesive and engaging content flow.
For the 1st and 3rd sections of the generated article, the prompt will be:
The article should tell about known facts reviving the secrets.
For the 2nd section, the prompt will be enhanced with additional directives:
The article should tell about known facts reviving the secrets.
Format the text for embedding in a WordPress post, using only <p>, <em>, <blockquote>, <ul>, <li> tags. Add one quote from your knowledge base. Exclude any other HTML tags.
For the 4th section, the prompt also contains its own specific instructions:
The article should tell about known facts reviving the secrets.
Format the text for embedding in a WordPress post, using only <p>, <em>, <table>, <tr>, <td> tags. Include one table with a useful information. Exclude any other HTML tags. Tell about a mystery or secret related to the topic and invite the reader to find the answer for themselves. Encourage them to think.
Note that the text sections of the generated article are numbered in regular order, starting from 1. This numbering helps in targeting and customizing each section with appropriate content through the use of sectionN parameters, ensuring that the content generation is organized and each section is precisely aligned with the given directives.
Avoid giving direct instructions to create an article in your prompts. The language model should understand that it is writing text that is part of an article. Otherwise, it may begin to generate sections of text that include an introduction and a conclusion, which are not desirable for this particular application. It’s important to ensure that the prompts are crafted in a way that directs the AI to produce content that fits seamlessly into the broader context of the article, without redundant structural elements. This approach helps maintain the coherence and focus of each section, optimizing the overall quality and relevance of the content.
The sectionN parameter gives you maximum control over the specific content of each text section in addition to the overall content of the article. Remember that you can use Spintax to further randomize the results, adding an element of uniqueness to each section generated. This attention to detail in the setup allows for a highly customizable output, tailored precisely to specific content goals and presentation styles.
Source: https://www.cyberseo.net/blog/controlling-the-content-of-text-sections-in-the-gpt_article-shortcode/
