aigc-video-cover-gpt SKILL.md skillA
aigc-video-cover-gpt from mr-salticidae/knowledge-base is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add mr-salticidae/knowledge-base/aigc-video-cover-gpt@v1.3Pins v1.3 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- mr-salticidae/knowledge-base
- Path
07_skill存档/aigc-video-cover-gpt/SKILL.md- Versions recorded
- 6
- Latest version
- v1.3, committed Sep 25, 2026
- First version recorded
- Jun 24, 2026
- Size
- 21,916 bytes
- Repository stars
- 33
- License
- a license file GitHub does not identify
用 GPT Image 2 一步生成「商单/科普视频」高质量横版封面的工作流。输入一张人物参考图 + 一份视频脚本，自动发散标题钩子、套用商单封面公式（构图不固定·暗 / 亮双影调随机）、写出含锁脸与中文大字策略的 GPT Image 2 提示词（中间产物），并一次出若干张可发布的 16:9 封面。当用户说"给这个视频做封面""根据脚本出封面""用 GPT Image 2 做视频封面""把人物参考图做成封面"时触发。.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# AIGC Video Cover (GPT Image 2) ## 适用场景 把「一张人物参考图 + 一份视频脚本」直接变成若干张可发布的 **16:9 横版视频封面**，由 **GPT Image 2 一步出含中文大标题的整张图**。面向商单视频、知识科普、揭秘解说这类需要「缩略图秒懂 + 有点击欲」的封面。 典型触发语： - "给这个视频做封面 / 出 3 张封面" - "根据脚本出封面，人物用这张参考图" - "用 GPT Image 2 把人物参考图做成封面" - "这条片子要发 B 站，封面帮我搞一下" **与相邻 skill 的区别**： - `aigc-poster-layout` 是**保护已定稿原图**做手工排版海报，不重绘主体；本 skill 是**从零生成**封面、接受 GPT Image 2 重画人物。 - `aigc-prompt-optimizer` 是通用 prompt 优化；本 skill 专做「脚本 → 封面」这一条龙，内置商单封面公式与锁脸打法。 ## 核心约定（本 skill 的硬设定） 1. **一步出整图**：GPT Image 2 直接生成含中文大标题的完整封面，不做「底图 + 后期排版」两段式。 2. **接受 ~92% AI 近似**：人物用参考图重画，靠「禁美化 + 点名锚点」把身份拽回到工具上限，不做抠图换脸。详见 [[GPTImage2锁脸的脸占比上限_v1]]。 3. **只出 16:9 横版**：用自然语言 `Wide 16:9 horizontal video thumbnail`，GPT Image 2 不吃 `--ar`。 4. **若干张 = 多个标题钩子**：默认从脚本发散 **3 个**不同钩子，各出 1 张，而非同一张改比例。 5. **构图不固定**：人物**不锁死在右侧**。从「构图库」（见第三步）随机选位——左置 / 居中 / 前景大头 / 底部探出 / 左右对峙 / 斜角 / 分屏 等都可；一批 3 张默认**构图各不相同**，让封面有变化、不千篇一律。唯一不变的是「文字与主体不互相糊住、主标题缩略图可读」。 6. **暗版 / 亮版双影调随机**：每张封面随机走「暗版」或「亮版」影调（见第三步）；一批 3 张默认**两种影调都出现**（如 2 暗 1 亮 / 1 暗 2 亮，随机），让用户在明暗两种风格里挑。 7. **prompt 是中间产物**：先把 prompt 写清楚给用户过目（尤其文字内容、锚点、构图、影调），再调用出图。 --- ## 第一步：拆参考图 + 拆脚本 ### 1A. 从参考图抽「身份锚点」 找出参考图**最强的单一特征**（眼睛颜色/形状、脸型、发型、肤色、标志性配饰其一），作为锁脸成败的判据。 …
Read the whole file at its exact version.
How to install
mdr add mr-salticidae/knowledge-base/aigc-video-cover-gpt@sha256:7fe93a6637c03998The line at the top of this page pins the label v1.3, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_zqxdxrdsvj3nkbg4)
4 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 6 versions of this file, between Jun 24, 2026 and Sep 25, 2026, about one every 19 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| v1.3 latest | 2026-09-25 | c9a2154 | 21,916 B | A | view · diff |
| v1.3 | 2026-06-26 | 5ff7203 | 21,913 B | A | view · diff |
| v1.2 | 2026-06-26 | 674f78a | 20,744 B | A | view · diff |
| v1.1 | 2026-06-26 | a5ea16d | 16,038 B | A | view · diff |
| v1 | 2026-06-24 | 010965f | 12,625 B | A | view · diff |
| v1 | 2026-06-24 | 8484f8b | 12,014 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (21916 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
mr-salticidae/knowledge-base · 33 stars · license a license file GitHub does not identify · pushed 2026-11-06 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_zqxdxrdsvj3nkbg4 GET https://markdownregistry.com/api/v1/resolve?ref=mr-salticidae/knowledge-base/aigc-video-cover-gpt GET https://markdownregistry.com/api/v1/blob/7fe93a6637c03998dda4d9570f757a70c7ec45c0f87641b8c41536c103420937
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.