logo-design-skill

logo-design-skill

Logo design skill for AI agents

AIOfficial site

Description

Ask an AI for a logo and you usually get an icon and a line of text stuck together, with no idea behind it, a blur at 16 pixels and nothing readable in one colour. logo-design-skill is a skill pack for Claude, Gemini CLI, Codex and other agents that teaches the AI a designer's whole process, from the first brief to delivery, so it builds marks that actually hold up.

It asks about the brief first, studies the category's conventions in a reference library to avoid look-alikes, writes 8 to 12 one-line ideas, builds three of them in SVG and tests each one. Then it stops at a checkpoint: the concepts on one overview sheet with a recommendation, and the full kit only after you pick a direction. The library holds more than 1,400 real SVG logos classified by mark type, technique and industry, used to study construction and never to copy.

Features



Complete design process: Brief, word mapping, mark type, concepting, geometric construction, optical corrections, colour, typography, lockups, testing, presentation and delivery, each with written principles.

Reference library: 1,400+ real SVG logos classified by mark type, technique, geometry, subject, typography, mood and industry, searchable from the command line and browsable in a local gallery.

Testing tools: Dependency-free Python scripts audit an SVG against logo principles and build 16 px, one-colour, reversed, squint, mirror, context and competitor shelf test sheets.

Presentation and delivery: Builds client presentation boards with industry mockups, renders PNGs, and exports a full favicon, app-icon and web-manifest set.

Concepts before the kit: It always stops at the checkpoint, so the most time-consuming kit work only happens for an approved idea.

Works across agents: It uses the open Agent Skills format. Install it in Claude Code from the plugin marketplace, or copy the folder into the skills directory of Gemini CLI, Codex CLI and others.