Develop Brand Mark

develop-brand-mark is a Codex skill for developing, critiquing, selecting, and productionizing distinctive brand marks.

It treats logo development as a governed visual process: research broadly, compare real visual territories, learn from rejection, lock a selected direction, construct canonical vector geometry, and validate the mark at the sizes and in the contexts where it will actually appear.

What the skill covers

• Brand and product brief formation
• Semiotic, category, geometry, and production research
• Genuinely different visual territories
• Comparable concept boards and actual-size proofs
• Accidental-reading audits
• Rejection-led resets
• Explicit direction locking
• Canonical SVG construction
• Responsive marks, favicons, icons, and lockups
• Motion concepts
• Light, dark, monochrome, masked, and small-size validation

The skill starts from a 16x16 survival requirement and distinguishes bitmap exploration from production vector authority.

Install

Clone the repository into your Codex skills directory:

git clone https://github.com/mikeylong/develop-brand-mark.git \
  ~/.codex/skills/develop-brand-mark

Restart Codex after installation so the skill is discovered.

Use

Invoke the skill explicitly:

$develop-brand-mark

Example:

Use $develop-brand-mark to explore and build a scalable identity mark for an
AI developer-tooling company. Start from the existing logo, show genuinely
different visual territories, and validate the selected direction at 16px.

The skill can begin from either a new brief or an existing logo and brand system. When an existing identity is supplied, it first inspects the governing assets and preserves selected geometry unless the user authorizes a reset.

Workflow

1. Establish the brief and visible risks.
2. Research visual territory and production constraints.
3. Explore multiple directions with comparable visuals.
4. Critique what each mark actually resembles.
5. Record rejected primitives and reset when necessary.
6. Lock the selected direction.
7. Build one canonical vector system and deliberate variants.
8. Render and inspect the real target sizes and contexts.
9. Deliver governed assets, proofs, and unresolved decisions.

Repository structure

.
├── SKILL.md
├── agents/
│   └── openai.yaml
└── references/
    └── selection-lock-template.md

• SKILL.md contains the complete workflow.
• agents/openai.yaml provides Codex interface metadata.
• references/selection-lock-template.md records the visual authority, canonical geometry, exclusions, treatments, and validation after selection.

Core principle

A coherent concept story is not proof of a successful visible mark. The mark must survive direct comparison, accidental-reading review, real rendering, and actual-size inspection.

License

No license has been specified yet. Public visibility does not grant permission to copy, modify, or redistribute the project until a license is added.