Image Gen
Write structured, style-aware prompts before calling an image model for app mockups, marketing art, or store assets.
Overview
Image-gen is an agent skill most often used in Build (also Launch distribution, Validate prototype) that teaches a five-part prompt structure and style keywords for reliable AI image generation.
Install
npx skills add https://github.com/marswaveai/skills --skill image-genWhat is this skill?
- Five-element prompt structure: subject, style, composition, lighting or mood, and quality modifiers
- Style keyword table covering eight treatments (photorealistic, cyberpunk, ink, watercolor, anime, oil, minimalist, vinta
- Worked examples from basic one-liners to advanced cinematic 8K-style prompts
- Composition tips for framing subjects and scene layout in generative requests
- Five-element prompt structure (subject, style, composition, lighting/mood, quality)
- Eight named style rows in the style keyword table
Adoption & trust: 855 installs on skills.sh; 64 GitHub stars; 1/3 security scanners passed (skills.sh audits).
What problem does it solve?
You know what picture you need but one-line prompts keep returning generic, off-brand, or low-detail images.
Who is it for?
Indie builders generating app UI concepts, social assets, or store graphics who want repeatable prompt quality without a design team.
Skip if: Teams that already have locked brand guidelines, fixed asset libraries, or no generative image step in their workflow.
When should I use this skill?
You are about to call an image model and need subject, style, composition, lighting, and quality modifiers spelled out.
What do I get? / Deliverables
You leave with a structured, style-tagged prompt ready to paste into your image model or automation step for mockups and marketing art.
- Structured natural-language image prompt
- Optional style-keyword line matched to your visual treatment
Recommended Skills
Journey fit
Spans multiple journey phases - primary shelf plus alternate fits below.
Visual assets are produced while designing product UI and brand surfaces, which is the canonical Build → frontend shelf even when prompts are reused later for launch creatives. Frontend and product-facing visuals (screens, icons, hero art) are the first concrete place solo builders need repeatable prompt recipes.
Where it fits
Draft a photorealistic mock of the core screen before committing to Figma or code.
Generate hero and empty-state illustrations that match your app's mood and composition rules.
Produce social and landing-page art with consistent lighting and quality modifiers.
How it compares
Prompt-pattern cheat sheet for image models, not an image API integration or MCP server.
Common Questions / FAQ
Who is image-gen for?
Solo and indie builders using Claude Code, Cursor, or Codex who call image models for product and marketing visuals and want consistent prompt structure.
When should I use image-gen?
Use it in Validate when sketching prototype visuals, in Build when drafting UI heroes and icons, and at Launch when creating distribution and social creatives—before you send the final string to your generator.
Is image-gen safe to install?
It is documentation-style guidance with no bundled executables; review the Security Audits panel on this Prism page before adding any repo skill to your agent.
SKILL.md
READMESKILL.md - Image Gen
# Image Prompt Guide ## Writing Good Prompts ### Structure A good image prompt has these elements (in any order): 1. **Subject**: What is in the image (person, object, scene) 2. **Style**: Art style or visual treatment 3. **Composition**: How elements are arranged 4. **Lighting/Mood**: Time of day, atmosphere 5. **Quality modifiers**: Detail level, rendering quality ### Examples **Basic**: "a cat sitting on a windowsill" **Better**: "a fluffy orange tabby cat sitting on a sunny windowsill, warm afternoon light, cozy interior, highly detailed, photorealistic" **Advanced**: "a fluffy orange tabby cat sitting on a vintage wooden windowsill, golden hour sunlight streaming through lace curtains, dust particles visible in light beams, bokeh background of a garden, photorealistic, 8K quality, cinematic composition" ## Style Keywords | Style | Keywords to add | |-------|----------------| | Photorealistic | photorealistic, highly detailed, 8K, professional photography | | Cyberpunk | neon lights, futuristic, dystopian, rain-slicked streets | | Ink painting | Chinese ink painting, traditional art style, brush strokes | | Watercolor | watercolor painting, soft edges, flowing colors | | Anime | anime style, Japanese animation, cel shading | | Oil painting | oil painting, thick brushstrokes, rich colors, canvas texture | | Minimalist | minimalist, clean lines, simple composition, white space | | Vintage | vintage, retro, film grain, muted colors, 1970s | ## Composition Tips - "close-up" / "portrait" for face/detail shots - "wide angle" / "panoramic" for landscapes - "top-down" / "bird's eye view" for overhead shots - "cinematic composition" for movie-like framing - "centered" / "rule of thirds" for specific placement ## Using Reference Images Reference images guide the AI on style, not content. Tips: - Use reference images for style transfer: "generate in this art style" - Two modes available: - **URL mode**: Direct image URLs (`.jpg`, `.png`, `.webp`, `.gif`) - **Local file mode**: Provide file paths — the agent encodes them as base64 (`.jpg`, `.png`, `.webp`, `.heic`, `.heif`) - Max 14 reference images per request - The prompt still controls the content; references control the visual style - For URL mode, recommended image hosts: imgbb.com, sm.ms, postimages.org, imgur.com ## Language Note Always write prompts in **English** — the image generation model is trained on English descriptions. If the user provides a Chinese prompt, translate it to English before submitting. --- name: image-gen description: | Generate AI images from text prompts. Triggers on: "生成图片", "画一张", "AI图", "generate image", "配图", "create picture", "draw", "visualize", "generate an image". metadata: openclaw: emoji: "🖼️" requires: bin: ["listenhub"] primaryBin: "listenhub" --- ## When to Use - User wants to generate an AI image from a text description - User says "generate image", "draw", "create picture", "配图" - User says "生成图片", "画一张", "AI图" - User needs a cover image, illustration, or concept art ## When NOT to Use - User wants to create audio content (use `/podcast`, `/speech`) - User wants to create a video (use `/explainer`) - User wants to edit an existing image (not supported) - User wants to extract content from a URL (use `/content-parser`) ## Purpose Generate AI images using the ListenHub CLI. Supports text prompts with optional reference images (local files or URLs), multiple resolutions, and aspect ratios. Images are saved as local files. ## Hard Constraints - Always check CLI auth following `shared/cli-authentication.md` - Follow `shared/cli-patterns.md` for command execution and error handling - Always read config following `shared/config-pattern.md` before any interaction - Output saved to `.listenhub/image-gen/YYYY-MM-DD-{jobId}/` — never `~/Downloads/` <HARD-GATE> Use the AskUserQuestion tool for every multiple-choice step — do NOT print options as plain text. Ask one question at a time. Wait f