Skip to content

4. The initial prompt to Gemini 3

We assume here that the reader has some prior knowledge of Gemini AI. If not, they are advised to read document [ref1]. The initial prompt given to Gemini was in the form of the following text file [Prompt User.txt]:

User requests

- Create a Python script [convert] to generate a MkDocs site using the Material theme from an ODT document. The name of the ODT document will be the sole argument for convert.

- Then create a Python script [build] that will use the [mkdocs.yml] file created by convert to generate a static HTML site
- Finally, the [build] script will display the sites homepage in a browser.

- The site will have two independent columns. The left column will contain the table of contents, and the right column will contain the content of the ODT document.

- The table of contents will include all the headings from the ODT document (Heading 1, Heading 2, Heading 3, etc.), their numbering, and their hierarchy

- The images are mostly contained within tables and have been resized.
You will search through the tables for these images and ensure their resized dimensions are preserved.

- There are code blocks in the ODT document. They can be identified by their styles [Numbered Source Code Results, Source Code]
Other attributes of the code paragraph should be ignored. For example, if the code paragraph is also a numbered list, you must ignore this attribute.
All these code blocks must be treated as code on the MkDocs site. The lines of code must be numbered in the HTML output.

- For all code blocks, you will use Python syntax highlighting.

- You must not alter the lines of code. In particular, you must not remove the indentations or spaces preceding the first character of the line of code.

- You must identify all bulleted lists. ODT has several ways to create them. You must explore them all. Bulleted lists are
sometimes nested within one another. You must preserve this nesting.

- The title of the site will be "Generate a Python script with AI tools"

- You will embed every image extracted from the ODT document as well as every MkDocs chapter created

- You must preserve the links found in the ODT document. They must remain as links in the generated HTML document.

- You must preserve text that is bold, underlined, or italicized. It must be rendered exactly as is in the HTML document
  • Line 19: The initial test was performed on a previously published course that contained Python code;
  • This prompt was used to start the chat. After a few dozen iterations, it became obsolete;