Create a permanent memory for project YAML conventions

Context

This prompt extends the permanent conventions and workflow for writing or editing YAML frontmatter in Markdown content files throughout the content library.

Why this matters:

Missing various properties in YAML frontmatter causes improper or undesired rendering, if not flat out errors, in rendering content in our content-driven application. Errors are a frequent cause of build failures and rendering bugs in content-driven projects. By working within these conventions, we ensure content is easy to maintain, automate, and scale—whether you’re a human contributor or an AI assistant.

Audience

  • Content creators, editors, and developers working with Markdown files in this project.
  • AI assistants or automation tools tasked with generating or updating YAML frontmatter.

Why this prompt:

We have created a successful process to generate images to go with our content.

Workflow

For Large Batches of Files: Use an LLM API or automation script to add or update YAML properties across many files at once. Always validate changes with a YAML linter before committing.
For Small Batches or Individual Files: Use the chat window in your tool (e.g., Cascade, VS Code, or GitHub Copilot Chat) to generate or edit YAML frontmatter.
While the AI Assistant is at it as a copywriter, always double-check for:
  • Proper opening and closing delimiters (---)
  • Correct indentation and syntax
  • Compare against a template if one exists, potentially noting required or suggested properties are present and valid.

Example:

content/lost-in-public/prompts
In the Prompt files, we asked an LLM Chat Client to review the files and to use creativity to write a short, vivid, descriptive "image_prompt" value as a string.
image_prompt: "A bold hero section UI with a striking headline, vibrant call-to-action button, and an engaging background image. The design is modern, clean, and visually impactful, drawing attention to the main message in a web interface."
We then ran a script to ping the Recraft API with that image prompt, and the script returns a "banner_image" value as a url.
We then use that banner_image url to render images in cards and other components that add life to content through images.

Task-at-Hand

Write an "image_prompt" for the files in the working directory. Make it one to four sentences, and try to use visual imagery that will help the Recraft API generate a meaningful, clear corresponding image to the specification.

Do not overwrite existing image_prompt values, do not remove any banner_image values.

Most files will not have an image_prompt value, but for those that do not, write one.

Working Directory:

content/prompts