describe-images-alt-text-no-markdown.yaml

tasks:
  - name: "DescribeImagesWithoutMarkdown"
    model: "models/gemini-1.5-flash"
    description: "An agent that describes each image in the input directory to help write alt text."
    preamble: "When describing an image, keep it short and concise, and focus on the essence of the image."
    steps:
      - prompt: "Provide a concise, descriptive alt text for this JPEG image that captures its essence and is suitable for users with visual impairments."
        flags:
          perfile: "<INPUT>"
          default_input: "./docs"
          file_ext: "jpg"
      - prompt: "Provide a concise, descriptive alt text for this PNG image that captures its essence and is suitable for users with visual impairments."
        flags:
          perfile: "<INPUT>"
          default_input: "./docs"
          file_ext: "png"
      - prompt: "Provide a concise, descriptive alt text for this GIF image that captures its essence and is suitable for users with visual impairments."
        flags:
          perfile: "<INPUT>"
          default_input: "./docs"
          file_ext: "gif"