AI Image Gen

AI图像生成与编辑。支持文生图、图+文生图、风格转换。当用户要求画图、生成图片、编辑图片、图片风格转换时使用此 skill。支持多种比例（1:1、3:2、16:9、21:9 等）和分辨率（标准、2K、4K）。

Security Scan

The skill's code and docs generally match an image-generation purpose, but there are incoherences (undeclared required env vars and non-official/default endpoints/affiliate referral) that could lead to accidental credential exposure — verify endpoints and provenance before use.

Purpose & Capability

The skill claims to perform AI image generation (Gemini Flash Image) and the included script implements that. However, the registry metadata declares no required environment variables or credentials while both SKILL.md and the script expect IMAGE_GEN_API_KEY and IMAGE_GEN_BASE_URL — a mismatch in declared requirements. Also the default BASE_URL (https://code.newcli.com/gemini) and README recommendation (FoxCode affiliate link) are not the official Google endpoints referenced by the 'Gemini' name, which is a provenance/clarity concern.

Instruction Scope

SKILL.md instructs the agent to set two env vars and run scripts/generate_image.py with a prompt and optional args. The runtime instructions are narrowly scoped to preparing a prompt, calling the API, and saving the returned image; they do not direct the agent to read unrelated files, system secrets, or other environment variables.

Install Mechanism

This is an instruction-only skill with one included script; there is no install spec, no downloads, and no archive extraction. Nothing is written to disk by an installer beyond the skill files already present.

Credentials

The script legitimately needs an API key and base URL for the image service, but those env vars are not listed in the registry metadata. The default base URL is a non-official host (code.newcli.com) and README pushes a third-party FoxCode service via an affiliate link; using the recommended flow could result in your API key and base URL being issued by or sent to a third party. The script sends the API key in a header (x-goog-api-key) to whatever BASE_URL is configured, so a malicious or misconfigured BASE_URL could receive your key.

Persistence & Privilege

The skill does not request persistent privileges (always:false), does not modify other skills or system-wide settings, and will not run autonomously beyond normal agent invocation controls.

What to consider before installing

Key points before installing or using this skill: - The script expects IMAGE_GEN_API_KEY and IMAGE_GEN_BASE_URL even though the registry metadata omitted them — you will need to supply these env vars. - The default/base URLs in the code and README are not the official Google endpoints; README promotes a third-party service (foxcode.rjj.cc) via an affiliate link. If you point IMAGE_GEN_BASE_URL to an untrusted host, your API key will be sent there (header x-goog-api-key). - If you intend to use a trusted provider, set IMAGE_GEN_BASE_URL explicitly to that provider's documented API endpoint. Do not use API keys issued by unknown middlemen unless you trust them. - Prefer testing with a restricted or disposable key and minimal permissions. Monitor network calls (or run in an isolated environment) until you confirm the endpoint behavior. - If provenance matters, ask the maintainer for the upstream/source of the backend (why it references 'Gemini' but uses code.newcli.com) and for an explanation of the FoxCode referral. If the author cannot explain or provide an official endpoint, treat the provider as untrusted.

AI 图像生成

通过执行脚本调用 Gemini Flash Image API 生成图片。

环境变量

脚本通过以下环境变量获取 API 配置：

• IMAGE_GEN_API_KEY — API 密钥
• IMAGE_GEN_BASE_URL — API 基础地址（默认：https://code.newcli.com/gemini）

使用方法

生成图片

执行脚本生成图片：

export IMAGE_GEN_API_KEY="your-api-key"
export IMAGE_GEN_BASE_URL="https://code.newcli.com/gemini"

python3 scripts/generate_image.py "你的提示词" --model gemini-3.1-flash-image-2k-16x9 --output output.png

参数说明

• prompt（必填）：图片描述提示词
• --model, -m：模型名称，默认 gemini-3.1-flash-image-2k-16x9
• --output, -o：输出文件路径，默认当前目录 generated_image.png

可用模型

默认模型：gemini-3.1-flash-image-2k-16x9（2K 分辨率，16:9 横屏）

基础模型（标准分辨率）

模型 ID	比例	适用场景
gemini-3.1-flash-image	1:1	社交媒体头像
gemini-3.1-flash-image-3x2	3:2	横版照片
gemini-3.1-flash-image-2x3	2:3	竖版海报
gemini-3.1-flash-image-3x4	3:4	竖版海报
gemini-3.1-flash-image-4x3	4:3	传统比例
gemini-3.1-flash-image-4x5	4:5	Instagram
gemini-3.1-flash-image-5x4	5:4	传统横版
gemini-3.1-flash-image-9x16	9:16	手机壁纸/短视频
gemini-3.1-flash-image-16x9	16:9	电脑壁纸/视频
gemini-3.1-flash-image-21x9	21:9	超宽屏

2K 分辨率（推荐）

模型 ID	比例
gemini-3.1-flash-image-2k	1:1
gemini-3.1-flash-image-2k-3x2	3:2
gemini-3.1-flash-image-2k-2x3	2:3
gemini-3.1-flash-image-2k-3x4	3:4
gemini-3.1-flash-image-2k-4x3	4:3
gemini-3.1-flash-image-2k-4x5	4:5
gemini-3.1-flash-image-2k-5x4	5:4
gemini-3.1-flash-image-2k-9x16	9:16
gemini-3.1-flash-image-2k-16x9	16:9
gemini-3.1-flash-image-2k-21x9	21:9

4K 分辨率

模型 ID	比例
gemini-3.1-flash-image-4k	1:1
gemini-3.1-flash-image-4k-3x2	3:2
gemini-3.1-flash-image-4k-2x3	2:3
gemini-3.1-flash-image-4k-3x4	3:4
gemini-3.1-flash-image-4k-4x3	4:3
gemini-3.1-flash-image-4k-4x5	4:5
gemini-3.1-flash-image-4k-5x4	5:4
gemini-3.1-flash-image-4k-9x16	9:16
gemini-3.1-flash-image-4k-16x9	16:9

工作流程

1. 理解需求：分析用户的图片需求，确定比例和分辨率
2. 选择模型：根据用途选择合适的模型（默认 2K-16x9）
3. 优化提示词：将用户描述扩展为详细的图片提示词
4. 执行脚本：调用 scripts/generate_image.py 生成图片
5. 展示结果：读取生成的图片展示给用户

提示词技巧

• 具体描述：主体、场景、光线、风格、色调
• 风格参考：水彩、油画、赛博朋克、吉卜力、写实摄影、中国工笔等
• 构图说明：视角（俯视/仰视）、景深、焦点
• 色彩指定：主色调、配色方案

示例

生成中国风后羿射日图：

python3 scripts/generate_image.py \
  "一位英勇的古代射手后羿站在险峻的高山之巅，身披金色战甲，肌肉紧绷，拉开巨大的神弓，一支燃烧着神圣光芒的箭矢对准天空中炽热的太阳。天空中有多个太阳同时燃烧，火焰倾泻而下。背景是翻滚的云海和龟裂的大地，远山在热浪中扭曲。中国传统工笔重彩风格，水墨与金箔点缀，色彩以深红、金色、墨黑、靛蓝为主，画面气势磅礴，具有敦煌壁画的厚重感。16:9 宽幅构图。" \
  --model gemini-3.1-flash-image-2k-16x9 \
  --output houyi.png