Brand Identity

Codex skill for generating project logos and icon packs from repository context using Gemini image models.

What It Does

• Reads code and documentation to extract brand signals.
• Produces a structured brand context and prompt directions.
• Generates logo concepts with Gemini (gemini-3-pro-image-preview by default).
• Exports production-ready assets:
  • favicon (16/32/48 + .ico)
  • extension icons (16/32/48/128)
  • high-resolution logos (256/512/1024)
  • web assets (apple-touch-icon, Android icons, OG image, web manifest)

Repository Structure

• SKILL.md: Core skill instructions and trigger description.
• agents/openai.yaml: Skill UI metadata.
• scripts/: Automation scripts.
• references/: Prompt framework, asset specs, and publishing checklist.
• LICENSE: MIT license text.

Requirements

• Python 3.10+
• Pillow for image packaging:
  • python -m pip install pillow
• Gemini API key:
  • GEMINI_API_KEY environment variable, or
  • --api-key CLI flag, or
  • interactive hidden prompt when not provided.

Quick Start

1. Extract repo context:

python scripts/extract_project_brand_context.py \
  --project-root . \
  --output-file brand-output/brand-context.md

2. Prepare:

• brand-brief.md
• logo-prompts.txt (prompt blocks separated by ---)

3. Generate concepts:

python scripts/generate_logos_gemini.py \
  --brief-file brand-brief.md \
  --prompt-file logo-prompts.txt \
  --output-dir brand-output/raw

4. Package selected concept:

python scripts/build_logo_assets.py \
  --input brand-output/raw/logo-01.png \
  --name my-product \
  --output-dir brand-output/final

Security Notes

• Never hardcode API keys in files.
• Do not commit .env or generated outputs containing secrets.

License

MIT. See LICENSE.

Created by Anshumani Ruddra.