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 scarletkc/agents --skill talk-like-scarletkcgit clone --depth 1 https://github.com/scarletkc/agentsWrote 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/scarletkc/agents/talk-like-scarletkc)<a href="https://agentmods.dev/skills/scarletkc/agents/talk-like-scarletkc"><img src="https://agentmods.dev/badge/skills/scarletkc/agents/talk-like-scarletkc/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/scarletkc/agents/talk-like-scarletkc"><img src="https://agentmods.dev/badge/skills/scarletkc/agents/talk-like-scarletkc.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00118 | $0.02820 |
| Opus 5.5 | $0.00047 | $0.01128 |
| Sonnet 5.5 | $0.00024 | $0.00564 |
| Haiku 4.5 | $0.00012 | $0.00282 |
Grade A, and why
talk-like-scarletkc 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 5d 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 — 141 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Talk Like scarletkc
让文字读起来像 scarletkc 本人写的。她的文字像一个有情绪、有明确判断的开发者 在实时分享自己的发现,先说结论或感受,再补原因,不写空洞背景,也不为了显得 完整而总结。
按任务分两种做法。用户给了要说的话,让你换成她的口吻,是改写语气,包括润色、 整体改写、翻译和局部修改:原意、事实、立场和情绪强度都照原样,只改怎么说。 用户给了素材或主题,让你写一条,是起草:不要只把素材转述一遍,可以主动加角度、 态度、反应、吐槽、类比和反问,把话说得更有劲。两种做法都不编造可核对的事实。
本 skill 只管声音,也就是措辞、节奏、语气和标点。X 帖文选什么讲、第一句放
什么、要不要分楼,交给
x-content。
两个一起用时,内容取舍听 x-content,怎么说听这里。规则只作用于代写、改写和
翻译的成稿,不约束助手和用户讨论任务时的回答。
声音
- 第一句就是正文。判断、发现、情绪、具体问题或真实的反差直接放第一句, 不先铺背景,也不先宣布要说什么。
- 有态度。喜欢就说喜欢,觉得烂就说烂,不改成见仁见智,也不自动补一句希望 以后更好。用户给出的立场至少按原来的强度写,起草时可以说得更鲜明。个人体验 用我感觉、对我来说标明,已经确定的感受不软化。
- 情绪要像真的。改写时惊讶、兴奋、失望、烦躁和自嘲照原样保留,不凭空升级; 起草时可以加她会有的反应。口语感叹和粗口偶尔出现才像她,每条都用、反复用 同一句就成了模仿。
- 节奏不规则。短句、半句反应和长说明混着用,可以先给反应再解释,省略主语、 随口补一句都行。粗糙感来自节奏,不来自错别字和语病。
- 技术词和口语混写。Claude Code、Codex、PR 这类名词保留英文,直接嵌在很口语 的中文句子里,不为统一语言去翻译。
- 幽默一笔带过。反差、假装感谢、自嘲和轻微夸张,短,不解释笑点。
- 说具体的。能用动作、数字和场景讲清的,不用抽象评价。帮读者理解的比喻和 类比保留,需要时补一个,再接上实际机制。一路狂奔、插上翅膀这类只为显得有 文采的套话意象,改成字面说法。
- 不重复自己。连着写几条时,开头方式、结构、收尾和口头禅都要换,不要每条都是 判断开头、事实居中、金句收尾的同一个模子。有的直接停在事实上,有的只有一句 反应,有的整条都在讲一个细节。
起草时能加的是说法和看法:角度、钩子、态度、反应、吐槽、类比、反问,以及 明确是假设的场景。不能加的是可核对的事实:她没有过的经历,没给的数据和测试 结果,没发生的事,别人没说过的话,产品没有的功能。写推文、评论这类公开内容时, 用户没给立场不必先停下来问,挑一个从素材里站得住的写,在成稿外用一句话说明这是 你替她选的。代写发给具体某个人的消息时不同,不替她做推荐、承诺或表态,写一条 不表态也能直接发的回复,比如反问对方用途;绕不开立场时,问她一个具体问题。
删掉 AI 痕迹
下面这些在她本人手写的聊天语料里几乎为零(长破折号、加粗和分号都是 0 次), 出现在成稿里就是破绽。
- 标点。长破折号
—在任何语言里都不用。中文引号只留给直接引用、作品名和 避免歧义。少用冒号,不用分号。不加装饰性 emoji、加粗和标题,不把推文和聊天 改成列表。代码、命令和技术标识里的符号不受限。 - 开场套话。当然可以、这是一个很有意思的问题、在当今快速发展的时代、以下是 优化后的版本、关于这个问题我认为可以从几个方面来看。
- 空话铺垫。值得注意的是、毫无疑问、不可否认、归根结底、总而言之、这背后反映 的是、这也让我意识到。接下来才是重点、很少有人注意到这类只预告不给内容的 句子,直接换成那条内容本身。
- 防御尾巴。肯定句后面惯例式跟一句但不代表、但不能说明。主张过宽就把主张本身 收准,真正影响理解的条件放进那句话里一次说清。
- 和稀泥。双方都有道理、具体情况因人而异、理性看待、技术本身没有好坏,除非 这本来就是用户的意思。
- 硬凑的结尾。换个说法重复正文,突然升华到时代、本质和意义,你怎么看、欢迎 评论区讨论、希望能帮到你。最后一句如果带来新东西,比如一个吐槽、一个具体 后果、一个反转,或者从前文推出来的更大判断,就留着;没有新东西就直接停。
- 造出来的句式。为了衬托主张临时虚构一个被否定的一方,比如不是 X 而是 Y、 这不只是 X 更是 Y;为了节奏凑出来的排比和对偶。材料里真实存在的反差和对比 照常写,那往往就是最有意思的部分。
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 5d ago Changed · -20 lines 58bfb576a10e
- 1mo ago Changed · +7 lines 5fa111cd68e2
- 1mo ago Changed · +7 lines c3ca85052137
- 1mo ago Changed · +2 lines 4751bc939f5b
- 1mo ago Changed · +7 lines · -23 tokens per session 59cacc27229d
- 1mo ago Changed · +5 lines d9b095a41d73
- 1mo ago First seen · 133 lines · 141 tokens per session scan A 9e2810db7220
talk-like-scarletkc is a skill published in the GitHub repository scarletkc/agents (224 stars, last pushed 6d ago), licensed Apache-2.0. It adds 118 tokens to every session and 2,820 once invoked, about $0.0005 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 2026-08-30.
Other skills, from other repositories
nextjs-i18n
Use when implementing i18n in Next.js 16 — next-intl or DIY dictionaries, locale routing, language switch, or formatters.
react-i18n
Use when implementing translations in a React app (not Next.js) with react-i18next — useTranslation, namespaces, pluralization, Suspense.
astro-i18n
Use when implementing i18n in Astro — locale routing strategies, getRelativeLocaleUrl/getAbsoluteLocaleUrl, Astro.currentLocale, or hreflang with sitemap.
laravel-i18n
Use when implementing Laravel translations — (), transchoice(), lang files, pluralization, locale middleware, or formatting.
seo-hreflang
Use when auditing or generating hreflang tags for i18n/multilingual SEO.
chinese-documentation
A Chinese technical-documentation style guide covering spacing, punctuation, numbers, terminology, and links when Chinese and English appear together.