This skill should be used when the user asks to "generate one image", "make a single image", "create a quick image", "generate an image quickly", "just need one image", or otherwise needs a single-shot image generation without session setup. Covers the minimal generate_image flow and when to skip the session machinery.
How this skill is triggered — by the user, by Claude, or both
Slash command
/openai-creative-plugin:openai-quickThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Generate a single image fast, without session overhead. Best for one-off images, prototypes, or exploration.
Generate a single image fast, without session overhead. Best for one-off images, prototypes, or exploration.
openai-sprite-series skill instead (sessions + images param)openai-ui-mockups skill insteadcontinue_editing after the first generationAsk the user what they want to generate (if not already described).
Check API status with get_configuration_status. If not configured, guide through setup:
OPENAI_API_KEY env variable (recommended), orconfigure_openai_token with their key from platform.openai.com/api-keysChoose parameters based on the request:
aspectRatio: 1:1 (square/social), 16:9 (wide/presentation), 9:16 (mobile/story), 3:2 (photo), 21:9 (cinematic)model: gpt-image-2 (default, best quality) — switch to gpt-image-1.5 only if transparent background or inputFidelity control is neededquality: low for quick exploration, medium for normal work, high for production outputGenerate with generate_image. Write a descriptive prompt:
If anchoring to an existing reference — add referenceImages: ["/absolute/path/to/ref.png"] (up to 16 refs, all paths must be valid — any invalid path is rejected before the request is sent):
gpt-image-2 (default): always high-fidelity, suitable for style anchoring; does NOT support inputFidelity param — omit itgpt-image-1.5: add inputFidelity: "high" when identity preservation matters (exact subject likeness); supports background: "transparent" if neededsend_creative_message instead (see openai-sprite-series or openai-workflows)Visually verify with Read tool. If the result needs tweaking, use continue_editing with a correction prompt.
Report back with the image path and any notable details about what was generated.
| Vague ❌ | Descriptive ✅ |
|---|---|
| "a cat" | "A ginger tabby cat sitting in a sunlit window, watercolor illustration, warm tones" |
| "a logo" | "Minimalist tech startup logo, letter 'G', geometric, dark blue on white, vector style" |
| "space scene" | "Deep space nebula, purple and gold clouds, distant stars, photorealistic, cinematic lighting" |
[Subject] + [Composition] + [Action/State] + [Location] + [Style] + [Technical specs]
Example: 2D pixel art wizard character [subject], centered, front-facing [composition], casting a fireball [action], plain white background [location], clean outlines, limited 16-color palette [style], 32x32 pixel grid [technical]
continue_editing("change X to Y") — targeted correctiongenerate_image — editing rarely fixes fundamental layoutcontinue_editingcontinue_editingWhen turnaround time matters more than final polish:
quality: "low" first for draft exploration — escalate to "medium" or "high" only for finalsjpeg renders faster than png when you don't need transparency — use png only when an alpha channel is required(Source: OpenAI cookbook)
See the openai-prompts skill for domain-specific prompt structures (sprites, UI, icons, characters).
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub 2901were/openai-creative-plugin --plugin openai-creative-plugin