Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add chenyuxiaojin/xiaochen-skills --skill cyxj-subtitle-wenangit clone --depth 1 https://github.com/chenyuxiaojin/xiaochen-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/chenyuxiaojin/xiaochen-skills/cyxj-subtitle-wenan)<a href="https://agentmods.dev/skills/chenyuxiaojin/xiaochen-skills/cyxj-subtitle-wenan"><img src="https://agentmods.dev/badge/skills/chenyuxiaojin/xiaochen-skills/cyxj-subtitle-wenan/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/chenyuxiaojin/xiaochen-skills/cyxj-subtitle-wenan"><img src="https://agentmods.dev/badge/skills/chenyuxiaojin/xiaochen-skills/cyxj-subtitle-wenan.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00141 | $0.01941 |
| Opus 5.5 | $0.00056 | $0.00776 |
| Sonnet 5.5 | $0.00028 | $0.00388 |
| Haiku 4.5 | $0.00014 | $0.00194 |
Grade A, and why
cyxj-subtitle-wenan scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.
字幕转文案(cyxj-subtitle-wenan)
把外文视频字幕变成可念的中文文案(逐字稿)。最终口播是照着文案念出来的,会带上念稿习惯;本 skill 负责的是文案本身。
第一原则:保原意。
听懂原片在说什么 → 用说得出口的中文写出来。表达习惯只作用于怎么说,不能改掉原片意思。
核心路径:听懂意思 → 按会怎么说写成文案(禁止:英文 → 书面中文 → 硬拆字幕行)。
公开 skill 对照(2026-09):外面没有完全对上的路径。吸收了「说得出口、别切成字幕、钩子与信息提炼」等规矩;不照搬 WPS 禁预览、默认分镜生图、「每句≤20 字」硬上限。
何时用 / 何时不用
用:
- 有 YouTube 链接、视频 ID、或已有
.srt/.vtt - 要去订阅催促、片尾导流、自我推销
- 要改成说得出口的中文文案 / 逐字稿
不用:
- 只要查逻辑断点 →
dbs-script-flow/cyxj-content - 只要优化开头 →
dbs-hook/cyxj-hook - 达芬奇成片 SRT 修正 →
cyxj-subfix - 逐字稿整理成文章 →
cyxj-transcript(方向相反:去口语化成文) - 用户明确要求「按我的观点重写 / 改成我的判断」→ 那是二次创作,先确认再写,并标明与原片的差异;默认不做
保原意(最高优先级)
- 先用一两句话写下原片核心论点(作者的判断)。
- 改写后的文案必须仍能对上这句论点;数据、因果、结论方向不得偷换。
- 可以删广告、压缩啰嗦、调语序、改成口语;不可以加原片没有的主线、发明原片没有的经历或机制、把相反观点或空安慰写进原片没有的位置。
- 正文默认忠实原意;未经用户点名,不做评论向重写。
自检:把文案给没看过原片的人听,其复述的中心思想是否还是原作者的?差很大 = 失败。
表达审美(只作用于「怎么说」)
写前读 灵感库/已发布/ 里 1~2 篇句式节奏接近的成片,学的是:
- 一口气说完一件事;修饰语挂名词上
- 忌字幕栈(半成品短行排队)
- 数据嵌进句子里念,少「第一点第二点」板书腔
- 少主持腔总起(「今天我们来讲清楚」)
- 长短句交错,不以「每句≤20 字」为硬上限
- 可选保留已发布里真实出现的语气词习惯(对吧 / 说实话 / 这一个),不臆造口头禅,不加戏改论点
写稿两原则(2026-10-01 · 只作用于「怎么说」,原意不动)
原则一:当面说测试。 每句都要是人和人当面聊天时真会说出口的样子。
- 写完每一句,想象当面跟朋友说这句话;朋友听到的是什么样,稿里就写成什么样。
- 用词取平常说话的完整说法,口语感来自句子完整,用词保持规整:「Anthropic 三天前发布了 Sonnet 5.5,在它发布页的最下边,有一个很小的脚注」。
- 引用前先垫一句人话:「我去找遍了官方文档,里面其实已经写着这样一句话……」
- 英文先翻成日常说法再上稿,比如 overthinking 说第三方用「容易想得过多」。
原则二:文案往动画展示靠。
- 写每一句时,让句子本身带一个能画出来的物件、数字或动作,做动画的人听一遍就知道画什么。
- 这一条只管句子怎么写。稿子里只有口播:不写分镜、不写画面说明、不写 🎬 框(2026-10-16 定)。画面到做动画阶段再单独出。
去 AI 味:
- 先过原则一的当面说测试,再对照已发布的说法密度与口语节奏
- 用户当面改的句子立刻固化
讨厌的表达(不是讨厌原片观点):
- 翻译腔、字幕栈、金句排比收束
- 通用短视频 Hook→Problem→Solution→CTA 模板腔
- 工整平均的 AI slop 句式
输入 / 输出
输入: 视频链接/ID、字幕文件、或纯文本;可选对照成稿、时长偏好。
输出: 灵感库/待发布/<标题>(逐字稿).md
Frontmatter:来源、作者、处理说明、status、created、风格对照(句式样本)、原片核心论点(一句话)。
正文以「逐字稿原文:」起头。
交付:路径 + 开场预览 + 删了哪些广告。
库路径未设置 CYXJ_VAULT_BASE 时按 ~/obsidian/灵感库 兜底,并向用户确认。
文案铁律
- 保原意 > 表达习惯 > 其它润色。
- 产物叫文案;口播是照着念。
- 去广告导流,保留原片数据与判断。
- 说得出口,不把长文切成字幕。
- 口语化重组,不是直译,也不是换题重写。
- 一条转写稿对准原片一个主论点(原片若有多个,按原片主线)。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 142 lines · 141 tokens per session scan A 46f7b81d1410
cyxj-subtitle-wenan is a skill published in the GitHub repository chenyuxiaojin/xiaochen-skills (8 stars, last pushed 5d ago), licensed MIT. It adds 141 tokens to every session and 1,941 once invoked, about $0.0006 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-17.
Other skills, from other repositories
add-localization
Adds or extends multilingual localization in a Power Pages code-site SPA. Use whenever the user asks to translate, localize, internationalize, support multiple languages, add a language selector, add locale routes, repair an i18n setup, or add languages to an existing React, Vue, Angular, or Astro Power Pages site.
hreflang-international
International / multilingual SEO audit focused on hreflang correctness: missing or broken return tags, wrong language/region codes, missing x-default, self-referencing errors, hreflang vs canonical conflicts, and inconsistent signals across HTML head, HTTP headers and XML sitemap. Also reviews URL structure (ccTLD vs…
adaptation
This skill should be used when the user asks to "make an audiobook", "narration script", "narrator", "ACX", "Findaway", "pronunciation guide", "how long is the audiobook", "adapt to a screenplay", "Fountain", "screenplay", "picture book", "spreads", "illustration notes", "art notes", "comic script", "comics script"…
content-translation
Translate and localize content for international SEO. Use when the user asks to "translate content", "localize my site", "multilingual SEO", "translate to ", "international SEO", "hreflang", "multi-language website", "global content strategy", or wants to expand to new markets.
triage
Work a repo's needs-triage backlog, and outside contributors' issues and pull requests, in batches the decider approves in one pass: each item gets a cheap check and one proposed move with a plain-words reason, and only approved moves are applied. Use when asked to triage or work through a needs-triage pile, or to…
retro
Read a sample of one repo's past agent sessions, Claude Code and Codex, find the friction the agents hit, and return a short ranked list of fixes in plain words, each labeled and routed to whoever owns the fix. Use when the user asks for a retro or retrospective on agent work, asks what keeps going wrong in a repo's…