thinkhard.SI

Prompt repository Writing & editing

Write alt text for the images in a page or document

You are publishing a web page, newsletter or report and need short, useful descriptions of each image for people who use screen readers.

Registry no.
WRT-024
Collection
Writing & editing
Laid out for
Any chat model
Score
10 out of 10
Blanks to fill
3
Added

Writing & editingAny chat

The prompt

Score 10 of 10, Excellent

Use it when You are publishing a web page, newsletter or report and need short, useful descriptions of each image for people who use screen readers.

  • {{where_published}}
  • {{page_topic}}
  • {{image_context}}
Full prompt and why it works
Prompt
Write alt text for the images I have attached, for {{where_published}}. Alt text is read aloud by screen readers to people who cannot see the images, so each description should give them what a sighted reader gets from the image in this context, not a list of everything in it.

What the page or document is about: {{page_topic}}

The images, in the order attached, each with the text that sits next to it (caption or surrounding paragraph):
"""
{{image_context}}
"""

For each image, give its number, a label of three to five words so I can match it, its type, and the alt text:
- Informative: one or two sentences, under 125 characters where you can, describing what matters for this page. Do not start with "Image of" or "Picture of", because screen readers already announce an image.
- Functional (a link or button): what happens when it is selected, such as "Download the price list".
- Decorative: write Empty alt text (alt="") so screen readers skip it.
- Complex (a chart, diagram or map): a short alt text naming it and its main point, plus a longer description of up to 80 words with the key figures, to place in the text or a linked description.
Quote any words that appear in an image exactly.

Format: one line per image, in the order attached, with the fields separated by a vertical bar. For example, the form for one informative image: 2 | Volunteers planting trees | informative | Four volunteers in hard hats planting a row of saplings beside a footpath.

Describe only what is visible. Do not guess people's names, roles, ages, ethnicity or feelings unless the surrounding text states them, and do not add facts from outside the image and its context. If an image is too small, blurred or ambiguous to describe accurately, or the number of images does not match my list, say so for that image instead of describing what it probably shows. I will check every description against its image before publishing.

Variables

{{where_published}}
Where the images will appear, for example "our charity's annual report web page".
{{page_topic}}
What the page or document is about, in a sentence.
{{image_context}}
For each image, in order: its caption or the sentence next to it, and whether it is a link. Attach the images themselves to the chat.

Why it works

  • Who it's for. Alt text is described as what a screen reader user needs from the image on this page, which keeps descriptions short and relevant rather than exhaustive.
  • Format. Each image is typed as informative, functional, decorative or complex, because each type needs a different kind of alt text, and decorative images need none.
  • What to avoid. It describes only what is visible and never guesses a person's identity, age or ethnicity, the details image descriptions most often get wrong.

How to use it

  1. Press Fill in variables and type your details into each blank. What you type stays in this browser.
  2. Copy the filled prompt into the AI tool you already use, such as ChatGPT, Claude or Gemini, or open it in the Grader or the Prompt Tester here.
  3. Before you rely on it, try it on three or four real examples. The Prompt Tester runs them all at once and shows which ones go wrong.
Use it from code

Get this prompt as JSON from the free REST API, with your own text in the blanks, or let an AI assistant fetch it through the MCP server. No key.

curl -s https://thinkhard.si/api/v1/prompts/alt-text-for-images

With the MCP server added to Claude, ChatGPT, Cursor or VS Code, call get_prompt with {"prompt": "alt-text-for-images"}, or just ask for it by name.