- Home
- Skills
- Prompt & Skill Engineering
- gpt image 2 prompt engine
gpt image 2 prompt engine
Structured prompt engineering for GPT-Image-2 using atomic schemas and industrial JSON templates for commercial design.
Free
Works with the AI Agents you already use
gpt image 2 prompt engine
Example session with this skill installed
Create an e-commerce hero image for a 30ml frosted glass perfume bottle with a gold cap. Minimalist white background, soft side lighting. Needs the text "Midnight Bloom" in a serif font. 3:4 ratio.
- Read your context and instructions
- Compiled the gpt image 2
I have structured your request using the ecommerce-hero template to ensure high-fidelity output and text accuracy.
GPT-Image-2 Structured Prompt
Category: Ecommerce Hero (ID: ecommerce-hero)
JSON Schema Configuration:
{
"type": "Ecommerce",
"product": {
"name": "30ml Perfume Bottle",
"material": "Frosted glass, metallic gold cap",
"color": "Translucent white and gold",
"key_feature": "Elegant textures and reflections"
},
"scene": "Minimalist white studio background, clean surface",
"composition": "Centered, 45-degree eye-level shot, shallow depth of field",
"lighting": "Soft side lighting, subtle shadows, high-end commercial photography",
"style": "Luxury, minimalist, 8K resolution",
"text_overlay": {
"headline": "Midnight Bloom",
"style": "Elegant Serif",
"position": "Lower third, centered"
},
"constraints": "Aspect ratio 3:4, text 'Midnight Bloom' must be spelled exactly and clearly readable, no watermarks, no distorted bottle shapes"
}
Final Generation Prompt:
(3:4) High-end commercial photography of a 30ml frosted glass perfume bottle with a gold cap, centered 45-degree angle, shallow depth of field. Minimalist white studio background, soft side lighting with elegant reflections. Text "Midnight Bloom" must be accurately displayed in an elegant serif font at the lower third. High resolution, 8K, no gibberish, no watermarks.
Next steps
- Copy the final prompt into your
image_generationtool to produce the first draft. - If the text is slightly blurred, re-run with the
reference_imageURL of the first result to refine details. - Adjust the
lightingfield in the JSON if you need a warmer or cooler mood.
Connects securely to your tools. The creator never sees your data.
About this skill
Generating high-quality, predictable visual assets for commercial use often results in inconsistent outputs and garbled text. Random prompting fails when you need specific compositions, readable typography, or strict brand adherence. This skill transforms vague image requests into structured, code-like prompts optimized for GPT-Image-2.
What it does
- Atomic Schema Decomposition breaks down requests into seven critical dimensions including Subject, Composition, Material, Typography, Lighting, Style, and Constraints.
- Industry Standard Templates provides 20+ JSON-based schemas for UI screenshots, infographics, movie posters, e-commerce hero shots, and brand identity assets.
- Structured Workflow executes a four-step process: category selection, case retrieval, template variable mapping, and iterative generation.
- Stability Enforcement applies five specific reliability techniques to lock in text accuracy, aspect ratios, and module counts.
How it works
- Match Category Identify the specific output type, such as
ui-screenshot-systemorecommerce-hero. - Query Reference Search the built-in library or remote raw repositories for successful case structures and compositions.
- Map Variables Populate the chosen JSON template with specific product details, color palettes, and text overlays.
- Generate & Refine Convert the schema into a final prompt string for the image generation tool and iterate based on visual feedback.
Frameworks & tools
This skill implements methodology from the awesome-gpt-image-2 repository. It utilizes Python scripts for template retrieval and integrates with standard image generation tools via natural language or JSON-to-text conversion.
Why this beats prompting it yourself
Manual prompting often overlooks negative constraints and specific layout instructions, leading to "hallucinated" text and messy compositions. This skill treats prompts as code, ensuring every visual element is defined, restricted, and repeatable across a product line or campaign.
Use cases
- Creating e-commerce product banners with accurate sales text and lighting.
- Generating technical infographics and knowledge cards with structured layouts.
- Designing high-fidelity App UI mockups for specific platforms like iOS or Android.
- Developing brand key visuals that maintain consistent materials and color palettes.
Known limitations
This skill is a prompt engine and does not generate images itself; it requires an external image generation tool. It is not designed for simple photography or artistic sketches where precise control is unnecessary.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Prompt & Skill Engineering.
Requires an external image generation tool. Not suitable for casual or non-structured artistic sketches. Repository access may fall back to raw URL fetching if local cloning is restricted.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
4 uses
Times used to date
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 23 days ago
- Free with an Agensi account