CommunityImagegithub.com

leviyuli/abstract_logo-skill

Distill a supplied image into a standalone abstract visual mark using source-specific geometry, flat inks, and subtle print texture. Use for image-derived abstract logos, photo-to-symbol requests, or abstract visual identity tasks.

What is abstract_logo-skill?

This skill transforms any supplied image into a unique, abstract visual mark—ideal for creating logos, symbols, or visual identities derived from photographs, sketches, or other imagery. It works by analyzing the source image for the two or three most distinguishing cues (such as silhouette, structural junction, gesture, or rhythm) and mapping those cues into a compact set of geometric modules, lines, overlaps, and negative-space shapes. The result is a self-contained, standalone mark that conveys character and composition without reproducing the original scene. The visual language is rooted in post-digital graphics, deconstructed Swiss layout, and diagrammatic abstraction, using flat spot-color inks extracted from the source, subtle print textures (like grain, photocopy noise, or registration drift), and precise geometry with restrained handmade variation. The skill supports both raster and editable vector outputs, and it checks the final mark at full and thumbnail sizes to ensure it reads as one coherent symbol. It excludes templated logos, commercial UI treatments, neon effects, and heavy aging, keeping the result refined and minimal. Designed for use with AI agents that can generate images or edit vectors, it provides a clear, repeatable process for translating visual content into abstract branding elements.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/leviyuli/abstract_logo-skill/tree/HEAD

Installed? Explore more Image skills: steipete/songsee, affaan-m/frontend-design, mattpocock/improve-codebase-architecture · View all 6 →

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Abstract Logo

Create one self-contained abstract mark from the supplied image. Ask for a source image if none is available.

Read and reduce the source

  1. Identify the two or three cues that most distinguish this image. Look for a silhouette, structural junction, gesture, repeated rhythm, or relationship between elements. Prefer cues that convey both character and composition.
  2. Map those cues to a few geometric modules, lines, cuts, overlaps, or negative-space shapes. Use the relationships between modules to carry meaning. Omit the complete figure, landscape, and incidental scene details. A different subject should call for a different construction.
  3. Give the mark the entire visual hierarchy. Center it or offset it slightly on a generous warm-white or gray-white ground. Keep it legible as a single mark at thumbnail size.

Set the visual language

  • Use post-digital graphics, deconstructed Swiss layout, generative symbol systems, and diagrammatic abstraction as design vocabulary. Derive alignment, repetition, or breaks in the grid from the source cues.
  • Add a sparse arc, partial construction grid, hairline, node, or micro-symbol only when it reveals something about the mark's construction or the source relationship. Keep those marks secondary. Exclude icon sets, infographic layouts, decorative patterns, and arbitrary geometry.
  • Extract one or two signature ink colors and one supporting ink color from the source. Assign each ink a role in separating modules or making a relationship clear. Use flat spot-color fills and restrained overprint where forms meet. Keep the palette compact and omit neon effects and commercial gradients.
  • Keep the geometry precise and let the print finish supply slight handmade variation. Use barely visible stencil grain, photocopy or scan noise, fine paper fibers, and localized registration drift. Leave ink edges subtly imperfect. Exclude heavy stains, cracks, and uniform aging.
  • Keep the result restrained. Exclude templated logo forms, commercial UI treatments, cyberpunk, exaggerated futurism, dense annotation, and watermarks.

Produce and review

For a finished raster graphic, use built-in image generation with the source as the visual reference. Describe the selected cues, how the forms express them, and the ink roles. If the user asks for an editable vector, build the mark from editable paths and keep any print texture separate from the core geometry.

Check the result at full and thumbnail size. Verify that every prominent shape serves a selected cue, the mark reads as one coherent symbol, and any construction details or texture stay subordinate. Deliver the standalone artwork. Include cue analysis only if requested.

Related Skills