video-agent-visual — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited video-agent-visual (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
美术师通过 GPTIMG2(gpt-image-2,OpenAI 兼容图片生成 API)批量生成 2K 图片素材:
storyboard.json,理解每个镜头的素材需求generate_images.py 批量生成图片storyboard.json(分镜师产出)visuals/*.png + visual-timeline.json + visual-report.mdpython scripts/generate_images.py <project_dir> [--style <风格>] [--concurrency <并发数>] [--aspect-ratio <比例>]参数:
project_dir — 项目目录(包含 storyboard.json)--style — 风格名称,对应 styles/ 目录下的文件(默认:default)--concurrency — 并发数(默认:5)--aspect-ratio — 宽高比(默认:16:9)脚本会:
storyboard.jsonis_post_production: true 的镜头/v1/images/generations 生成 2K 图片(response_format=url,拿到 url 后下载落地)visuals/ 目录(001.png, 002.png...)visual-report.md 和 visual-timeline.json--aspect-ratio 按下表映射到 GPTIMG2 的 2K 尺寸(边长对齐 16 的倍数):
| 宽高比 | 尺寸 |
|---|---|
| 16:9(默认) | 2560x1440 |
| 9:16 | 1440x2560 |
| 1:1 | 2048x2048 |
| 4:3 | 2048x1536 |
| 3:4 | 1536x2048 |
未列出的比例兜底为 16:9(2560x1440)。当前 storyboard.json 结构不含比例字段,比例由 --aspect-ratio 参数决定。
预置风格文件位于 styles/ 目录:
| 文件 | 说明 | 适用场景 |
|---|---|---|
default.txt | 写实、电影感、自然光 | 通用 |
tech.txt | 未来感、蓝色调、科技元素 | AI / 科技类视频 |
knowledge.txt | 温暖、清晰、学术感 | 认知 / 知识类视频 |
用户可新增自定义风格文件到 styles/ 目录,脚本会自动识别。
{
"video_specs": {
"resolution": "1920x1080",
"fps": 30,
"aspect_ratio": "16:9"
},
"total_duration": 490.0,
"visuals": [
{
"shot": "001",
"file": "visuals/001.png",
"description": "A person scrolling through phone...",
"start_time": 0.0,
"end_time": 4.0,
"duration": 4.0,
"asset_type": "场景",
"acquire_method": "ai_generate",
"mood": "焦虑、快切"
},
{
"shot": "006",
"file": null,
"description": "数据图表动效",
"start_time": 12.0,
"end_time": 14.0,
"duration": 2.0,
"asset_type": "数据",
"acquire_method": "post_production",
"mood": "冲击、停顿",
"editor_note": "数据图表动效"
}
]
}# 素材生成报告
## 统计
- 总镜头数:50
- 生成成功:35
- 生成失败:2
- 后期制作(跳过):13
## 明细
| 镜头 | 状态 | 文件 | 备注 |
|------|------|------|------|
| 001 | ✅ | visuals/001.png | |
| 006 | 📋 | - | 后期制作 (数据) |GPTIMG2(gpt-image-2)走 OpenAI 兼容 HTTP:POST {GPTIMG2_BASE_URL}/v1/images/generations, 鉴权头 Authorization: Bearer {GPTIMG2_API_KEY},请求体 response_format=url,从响应 data[0].url 下载图片落地。
| 环境变量 | 用途 |
|---|---|
| GPTIMG2_BASE_URL | GPTIMG2 服务基址(如 https://api.chatgpt-code.com,末尾不带 `/v1`,脚本自动拼 /v1/images/generations) |
| GPTIMG2_API_KEY | GPTIMG2 API 密钥 |
环境变量优先;缺失时脚本从 ~/项目/自己的应用/密钥存储/.env 解析同名键。
scripts/generate_images.py — 批量图片生成主脚本styles/default.txt — 默认风格指令styles/tech.txt — AI 科技类风格styles/knowledge.txt — 认知/知识类风格~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.