Expert techniques for crafting high-quality, detailed prompts for image generation and editing models. Covers composition, style, lighting, iteration strategies, and refinement workflows.
SkillScore82Maintained
Automatic
◆AI Engineer pack
Last verified 8/6/2026
What it changes
Passed 18 of 22 test cases with the skill, and 13 of 22 without it.
Without skill59%
With skill82%
Checks it passed that the model alone missed
Asks targeted questions about vehicle type, setting, era, and aesthetic mood
Asks clarifying questions to narrow down specific visual details or lighting nuances
Prompts the user to share specific feedback on generated results for continuous iteration
and 3 more in the full report
6 fixed12 passed either way3 still failing1 broke
Pass-rate lift
+23 pp
Tokens
+1%
Agent turns
0%
Response time
-15%
Model gemini-3.6-flashMeasured 8/6/2026v118 of 22 cases passed with skill
Tokens, turns and time describe resource use, not a dollar price.
See it in action
One of 6 test cases the skill fixed. Same model, same prompt, with and without the skill.
1 of 6
Prompt
My previous image prompt for a futuristic cybernetic city turned out flat and washed out. I want to refine it so the atmospheric lighting is much bolder and the composition uses a low-angle perspective, while keeping the rainy street scene intact. How should I revise my prompt to achieve a more cinematic look?
What was checked, without and with the skill
Checked
Without skill
With skill
Preserves existing core elements such as the rainy street setting while updating lighting and camera angle
Passed
Passed
Asks clarifying questions or prompts for feedback on specific visual aspects before finalizing next steps
Missed
Passed
Includes specific technical and lighting terms like low-angle wide shot and dramatic contrast
Passed
Passed
case-02 · gemini-3.6-flash · 8/6/2026 · answers are capped and redacted
What's inside
The instructions your model reads when the skill loads.
SKILL.md32 linessingle file
Image Prompt Expert
When active, help users create precise, effective image generation prompts and guide iterative refinement.
Core Principles
Specificity wins: Include subject, action, environment, lighting, mood, camera angle, style references, technical quality.
Structure prompts logically (main subject first, then details, then style/technical).
Support iteration: Generate, then refine based on results ("make the lighting more dramatic", "change style to cinematic").
Use references: Mention artists, art movements, films, or specific aesthetics when appropriate.
Prompt Structure Template
"[Subject/action in scene], [environment/setting], [lighting and atmosphere], [mood/emotion], [camera angle or composition], in the style of [artist/film/movement], [technical: highly detailed, 8k, sharp focus, cinematic lighting]"
Best Practices
Ask clarifying questions about desired style, mood, or references before drafting.
Provide 2-3 prompt variations (different emphases).
For editing: Give precise change instructions while preserving good elements.
Encourage user to describe what they like/dislike in generated images for next iteration.
Common Enhancers
Lighting: cinematic, dramatic, golden hour, neon, soft diffused
Composition: rule of thirds, leading lines, close-up, wide shot, aerial
Quality boosters: intricate details, sharp focus, masterpiece, best quality