nuyoah-ai-works/nuyoah-portrait-character-designer

State the character’s temperament, facial features, or makeup you want, and you’ll get a character direction and a complete Chinese prompt. Normally, “Design a character” outputs text first; Only after explicitly saying “generate image” does the current tool’s image capabilities be invoked.

Download

• Download the full ZIP version of v0.2.1
• Check versions and updates
• Read real cases and complete prompts

Please download the entire folder and keep the accompanying files such as references/. Copying only SKILL.md will miss the design rules it references.

Install in Codex

Send the following passage to Codex, which has local file capabilities:

After installation, enter the example below in the next round. If not recognized, first check the installation results and file location, then restart Codex.

You can also install it manually: download and extract the ZIP, then place the entire nuyoah-portrait-character-designer folder into your personal directory ~/.agents/skills/. You should eventually find ~/.agents/skills/nuyoah-portrait-character-designer/SKILL.md, but don’t add another folder. If you already have a version with the same name, keep your own modifications for now. For the location and discovery mechanism, see OpenAI’s official notes.

This is a standalone Agent Skill, not a plugin already listed in the plugin directory. For other tools supporting Agent Skills, please use their respective installation portals; This version of the use case is from Codex and does not guarantee consistent performance among all hosts.

First time using it

Normally, you’ll get a character direction and a complete Chinese prompt that can be copied separately. You don’t need to fill out forms yourself or learn bone structure terminology first.

Once you determine the direction, you can continue:

Image generation requires the current host to have image capability; the amount or fee is calculated based on the platform used. If you don’t have an image tool, you can also copy the full prompt into your own image generation tool.

Continue modifying

• “I want to design a Japanese-style Showa-era vibe for adult female characters.”
• “The same set of peach-pink makeup, designing five adult characters with different faces.”
• “Keep the character’s features and skin tone, just soften the eyeliner a bit.”
• “This image only removes makeup; don’t bring the reference person’s face over.”
• “Only give prompts.”

Local modifications return the complete draft by default; Multiple characters are given complete prompts separately. When changing the fixed face, the original structure is retained; When changing the face with fixed makeup, the main color, texture, line direction, and relative placement are retained.

Design Methods and Boundaries

Original characters will depict the specific relationships between the brow bones/eye sockets, cheekbones, and jaw/chin, and separate bone structure, cheek soft tissue, makeup, and lighting. Soft bone structure is also an effective design; Do not bind any particular temperament, nationality, or skin tone to a single face shape. For detailed candidates, see the word list.

This skill focuses on character structure and makeup. Complete photo planning, general poster reverse, and video cross-camera continuity require other skills; this skill does not require you to install those nearby tools.

Reference images only describe visible information and cannot infer unknown bone structure from strong side lighting or occlusion. Text design for different faces and retaining character instructions do not necessarily mean that the actual raw image will maintain identity or meet the expected aesthetic every time.

Versions and Validation

v0.2.