Claude Skill

direct-image-creation

Turn a creative request into a production-ready AI image brief, reference plan, node workflow, and model-profile prompt. Use for text-to-image, surgical editing, exact text or infographics, image analysis and style transfer, logos and brand kits, ads and publishing assets, ecomme

LLM Mart · 0 points · 0 views 0 listing impressions 0 install-command copies
Download maomaochong-ai-open-astravia-packages_plugins_presets_content-creation_agent_skills_direct-image-creation-cf58759.zip · 28 KB

Install

skills CLI npx skills add https://github.com/maomaochong-ai/open-astravia/tree/main/packages/plugins/presets/content-creation/agent/skills/direct-image-creation
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install maomaochong-ai-open-astravia@llmmart
Git git clone https://github.com/maomaochong-ai/open-astravia.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole maomaochong-ai/open-astravia collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Direct AI image creation

Use $operate-content-workflow for inspection and mutations. Capabilities are the source of truth for executable models and modes.

Route the task

  1. Inspect project state, references, and image capabilities.
  2. If no concrete visual concept exists, invoke $develop-creative-concept before authoring prompts.
  3. Classify the request as new generation, edit, variation, composition transfer, character/product continuity, or multi-asset set.
  4. Define the brief and acceptance criteria before creating nodes.
  5. Generate a small candidate set when direction is uncertain; select before producing expensive derivatives.
  6. Review the actual output with $review-content-quality and repair the smallest failing dimension.

Read only the references needed:

Brief requirements

Record purpose, audience, publishing surface, aspect ratio, subject, action/pose, environment, composition, lighting, palette, medium, references and their roles, immutable details, exclusions, deliverables, and acceptance criteria. Distinguish hard constraints from preferences.

Prefer an explicit visual decision over a pile of adjectives. Make each variation change one named axis such as composition, palette, lens, pose, or rendering medium.

This method is an original Astravia adaptation informed by Generative-Media-Skills (MIT), visual-skills by Serge Shima (CC BY 4.0, https://github.com/smixs/visual-skills), and ViMax (MIT).

Files (open-astravia)
  • agents
    • openai.yaml 250 B
      Loading…
  • references
    • brand-and-publishing-recipes.md 5.3 KB
      Loading…
    • commerce-and-food-patterns.md 2 KB
      Loading…
    • commerce-product-and-spatial-recipes.md 4.2 KB
      Loading…
    • editing-and-continuity.md 1.6 KB
      Loading…
    • fashion-portrait-and-character-patterns.md 2.1 KB
      Loading…
    • identity-fashion-and-social-effect-recipes.md 4 KB
      Loading…
    • interface-storyboard-and-layout-recipes.md 3.5 KB
      Loading…
    • model-prompt-profiles.md 2.6 KB
      Loading…
    • model-routing.md 1.2 KB
      Loading…
    • multi-panel-and-sequential.md 1.8 KB
      Loading…
    • poster-ui-and-social-patterns.md 2 KB
      Loading…
    • presentation-visuals.md 1.5 KB
      Loading…
    • production-prompt-skeletons.md 2.8 KB
      Loading…
    • prompt-framework.md 1.8 KB
      Loading…
    • quality-checklist.md 1.1 KB
      Loading…
    • scenario-routing.md 2.8 KB
      Loading…
    • structural-and-dimensional-control.md 1.8 KB
      Loading…
    • task-patterns.md 1.4 KB
      Loading…
    • text-and-information-design.md 2 KB
      Loading…
    • visual-decomposition.md 1.8 KB
      Loading…
  • SKILL.md 4.7 KB
    ---
    name: direct-image-creation
    description: Turn a creative request into a production-ready AI image brief, reference plan, node workflow, and model-profile prompt. Use for text-to-image, surgical editing, exact text or infographics, image analysis and style transfer, logos and brand kits, ads and publishing assets, ecommerce listing sets, product/food/fashion/portrait imagery, virtual try-on and identity-preserving effects, interiors and floor plans, UI mockups, character sheets, presentation visuals, social covers and thumbnails, multi-panel grids, storyboards, visual variants, composition changes, or improving an image-generation node.
    ---
    
    # Direct AI image creation
    
    Use `$operate-content-workflow` for inspection and mutations. Capabilities are the source of truth for executable models and modes.
    
    ## Route the task
    
    1. Inspect project state, references, and image capabilities.
    2. If no concrete visual concept exists, invoke `$develop-creative-concept` before authoring prompts.
    3. Classify the request as new generation, edit, variation, composition transfer, character/product continuity, or multi-asset set.
    4. Define the brief and acceptance criteria before creating nodes.
    5. Generate a small candidate set when direction is uncertain; select before producing expensive derivatives.
    6. Review the actual output with `$review-content-quality` and repair the smallest failing dimension.
    
    Read only the references needed:
    
    - Any prompt: read [references/model-prompt-profiles.md](references/model-prompt-profiles.md), then [references/prompt-framework.md](references/prompt-framework.md)
    - Ready-to-fill prompt structures: [references/production-prompt-skeletons.md](references/production-prompt-skeletons.md)
    - Editing or reference continuity: [references/editing-and-continuity.md](references/editing-and-continuity.md)
    - Model/mode choice: [references/model-routing.md](references/model-routing.md)
    - Exact text, infographic, diagram, or localization: [references/text-and-information-design.md](references/text-and-information-design.md)
    - Recreate, analyze, or transfer an attached visual: [references/visual-decomposition.md](references/visual-decomposition.md)
    - Sketch, wireframe, floor plan, or 2D/3D translation: [references/structural-and-dimensional-control.md](references/structural-and-dimensional-control.md)
    - Grid, collage, contact sheet, comic, or storyboard sheet: [references/multi-panel-and-sequential.md](references/multi-panel-and-sequential.md)
    - Slide visual: [references/presentation-visuals.md](references/presentation-visuals.md)
    - Product, character, storyboard, or social patterns: [references/task-patterns.md](references/task-patterns.md)
    - Commerce, product, food, or beverage: [references/commerce-and-food-patterns.md](references/commerce-and-food-patterns.md)
    - Fashion, portrait, or character design: [references/fashion-portrait-and-character-patterns.md](references/fashion-portrait-and-character-patterns.md)
    - Poster, illustration, UI mockup, or social asset: [references/poster-ui-and-social-patterns.md](references/poster-ui-and-social-patterns.md)
    - Need to choose a specialized production recipe: [references/scenario-routing.md](references/scenario-routing.md)
    - Logo, brand system, advertising, brochure, blog, social cover, or thumbnail: [references/brand-and-publishing-recipes.md](references/brand-and-publishing-recipes.md)
    - Ecommerce listing, product angle set, spatial design, floor plan, or interior visualization: [references/commerce-product-and-spatial-recipes.md](references/commerce-product-and-spatial-recipes.md)
    - Identity-preserving portrait pack, virtual try-on, color board, collectible, collage, or selfie effect: [references/identity-fashion-and-social-effect-recipes.md](references/identity-fashion-and-social-effect-recipes.md)
    - UI redesign, URL-to-design, dense information board, text object, or storyboard layout: [references/interface-storyboard-and-layout-recipes.md](references/interface-storyboard-and-layout-recipes.md)
    - Output review: [references/quality-checklist.md](references/quality-checklist.md)
    
    ## Brief requirements
    
    Record purpose, audience, publishing surface, aspect ratio, subject, action/pose, environment, composition, lighting, palette, medium, references and their roles, immutable details, exclusions, deliverables, and acceptance criteria. Distinguish hard constraints from preferences.
    
    Prefer an explicit visual decision over a pile of adjectives. Make each variation change one named axis such as composition, palette, lens, pose, or rendering medium.
    
    This method is an original Astravia adaptation informed by Generative-Media-Skills (MIT), visual-skills by Serge Shima (CC BY 4.0, https://github.com/smixs/visual-skills), and ViMax (MIT).
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.