thinkhard.SI

Prompt repository Image & video

Create a social media graphic with readable text

You need a post image with a short headline and your colors, sized for a specific platform.

Registry no.
MED-007
Collection
Image & video
Laid out for
GPT
Score
10 out of 10
Blanks to fill
9
Added

Image & videoGPT

The prompt

Score 10 of 10, Excellent

Use it when You need a post image with a short headline and your colors, sized for a specific platform.

  • {{platform}}
  • {{topic}}
  • {{audience}}
  • {{visual}}
  • {{style}}
  • {{brand_colors}}
  • {{headline}}
  • {{supporting_text}}
  • {{aspect_ratio}}
Full prompt and why it works
Prompt
# Role
You are a graphic designer making one social media graphic.

# Goal
Create one image for a {{platform}} post about {{topic}}, aimed at {{audience}}. It must read at a glance on a phone screen in a scrolling feed.

# The image
The single focal point is {{visual}}, placed in the lower half of the frame. The headline sits in the upper half on a plain, flat area of one color from the palette, so nothing busy sits behind the words. The style is {{style}}. Use only these colors: {{brand_colors}}, with the strongest one kept for the focal point or the headline. The lighting is soft and even, and the composition is simple, with generous empty space around the text. If I attach one of our past posts, treat it as the style reference: match its typeface, colors and layout, but not its words.

# Text on the image
Render exactly this text and no other words:
- Headline: "{{headline}}", large, in a bold, clean sans-serif typeface, on at most two lines.
- Supporting line: "{{supporting_text}}", much smaller, directly below the headline.
Spell every word exactly as written, with the same capitalization and punctuation.

# Success criteria
- The headline can be read at thumbnail size, with strong contrast between text and background.
- Someone scrolling understands what the post is about within two seconds.
- All text and key elements sit at least a tenth of the width away from every edge, so the image can be cropped and the app's buttons cover nothing.

# Constraints
- Leave the bottom-right corner empty, about a fifth of the width, where I will add our logo afterward. Do not draw a logo; generated logos come out distorted.
- No other text, watermarks, interface elements such as fake buttons or play icons, real people or recognizable characters.

# Output
One image with an aspect ratio of {{aspect_ratio}}. If you cannot make that exact shape, use the closest one you can and keep everything important well inside the frame, so I can crop it to size.

# Stop rules
If the supporting line is "none", leave it out. If the text will not fit at a readable size, ask me before shortening it; never change the words to make them fit.

Variables

{{platform}}
Where it will be posted, for example "Instagram feed" or "LinkedIn".
{{topic}}
What the post is about.
{{audience}}
Who the post is for.
{{visual}}
The main picture, concretely, for example "a flat illustration of three seed packets fanned out on a wooden table".
{{style}}
The look, for example "flat vector illustration with no outlines" or "bright, clean product photograph".
{{brand_colors}}
Your colors by name and hex code, for example "forest green #1F5130, cream #F4EFE6, coral #E4664A".
{{headline}}
The exact headline, ideally under 8 words. Shorter text renders more reliably.
{{supporting_text}}
One short supporting line, exactly as it should appear, or "none".
{{aspect_ratio}}
The shape for the platform, for example 4:5 for an Instagram feed post or 1:1 for a square post. Check the platform's current size guide.

Why it works

  • Format. Exact text in quotation marks, with its size and place in the image, is how OpenAI recommends asking for rendered text, and the spelling rule covers the most common failure.
  • What to avoid. An empty corner for your real logo, added afterward, avoids the distorted logos image models tend to draw.
  • If unsure. If the words will not fit, it asks before shortening them, and a safe margin lets you crop the image to the platform's exact size.

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/social-graphic-with-text

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