按 scarletkc 本人的自然表达习惯代写、改写、润色和翻译文本，适用于推文、评论、聊天消息、模型或工具体验文、项目介绍、GitHub 文本和正式通信。用户要求撰写可直接使用的成稿、去除 AI 腔，或在翻译中保留本人语气和立场时使用，无需明确点名本 skill。单纯的事实问答、技术分析、代码审查和任务讨论不触发。
The English summary is being prepared. Showing the original description.
Review the instructions the agent will receive before you install.
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。
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。
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。
— 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。英文同理，不写 corporate English，避开 game-changer、seamless、delve、 It's not just X, it's Y、Hope this helps 这类说法。日语不堆敬语模板。繁体中文 只做用字转换，诗音写作詩音。
同一个声音在不同场景松紧不同。按成稿去哪里判断：发给不特定的人看是 Social， 一对一即时交流是 Chat，进仓库是 Project writing，走正式流程是 Formal。
| 场景 | 适用 | 要点 |
|---|---|---|
| Chat | 聊天消息、给 AI 下指令 | 最松。跟人聊天和给 AI 下指令差别很大，写前读 references/surface-profiles.md 的 Chat 一节 |
| Social | X、微博、评论、公开短文 | 见下面的推文一节 |
| Technical opinion | 模型评价、开发工具体验、架构观点 | 先说实际体验或结论，再用具体功能、例子或工作流说明好在哪、差在哪。可以直接比较，不假装测评机构，事实、推测和感受分开 |
| Project writing | README、PR、changelog、issue、技术报告 | 读 references/surface-profiles.md 对应一节 |
| Formal | 客服邮件、申诉、正式回复 | 简短、礼貌、请求明确，不寒暄、不反复感谢、没有网络口语。三句话说清诉求比套满敬语更像她 |
| Translation | 翻译 | 保留情绪强度、立场、直接程度、节奏和幽默。翻译改变语言，不改变人 |
推文默认偏口语，不主动提高原稿的正式程度。她反馈过话术和标点太标准，要一起 改措辞、句子和停顿，只删几个句号不够。
references/dialogue-samples.md 校准长度、断句和连发拆分。scripts/lint_style.py 可以找候选，提示不等于必须改。| 文件 | 内容 | 什么时候读 |
|---|---|---|
references/surface-profiles.md | Chat 和 Project writing（含技术报告）的细则与语料统计 | 写这两类场景时 |
references/dialogue-samples.md | 100 段真实一对一对话，保留连发拆分 | 代写跟人聊天的消息时 |
references/persona.md | 身份背景和使用边界 | 涉及署名、人称或身份时 |
scripts/lint_style.py | 风格候选提示，只提示不改写 | 交付前可选 |
规范版本在 https://github.com/scarletkc/agents 的 skills/talk-like-scarletkc/。
用户纠正过的输出提炼成规则，不收集成推文样本；加规则前先看现有的一条能不能
改得覆盖它，能改一条就不新加一条。本地副本里的改动建议提 PR 回去。
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Delegate bounded investigation, review, or implementation tasks to Antigravity CLI from a supervising agent. Includes a background client for task progress, follow-up prompts, conversation recovery, cancellation, and visible failures.
The English summary is being prepared. Showing the original description.
Guide a user from an unclear idea to an actionable plan through the agent harness's native ask tool and clickable choices. Use when the user wants a guided requirements interview, step-by-step questions, help deciding what to build, or a button-led path from goals to scope, solution, and technology choices. Do not turn an ordinary implementation request or a single clarification into an interview.
The English summary is being prepared. Showing the original description.
When handing work to the Codex CLI earns its cost, and how to size the run: second-model review, bounded implementation hand-offs, sandbox permissions, model and reasoning effort. Use when the user asks for Codex or `codex exec`, when a change is complex or high-stakes enough that an independent reviewer would change the outcome, or when a delegated run needs its model, effort, or permissions chosen. For agents other than Codex itself.
The English summary is being prepared. Showing the original description.
Delegate bounded coding tasks to the local Devin CLI over ACP from a supervising agent such as Claude Code or Codex. Use when the user asks to hand work to Devin, when an independent implementation, port, review, or investigation can run in parallel with your own work, or when an earlier Devin session needs a follow-up. Includes a blocking run helper with single-use permission answers, a deny policy, progress summaries, and session resume.
The English summary is being prepared. Showing the original description.
Delegate bounded tasks to Grok Build from a supervising agent such as Codex. Use when the user asks Grok to investigate, review, implement, or provide another perspective, or when a distinct coding agent would materially help. Includes a background ACP client for progress, follow-up prompts, permissions, and cancellation.
The English summary is being prepared. Showing the original description.
Write outbound promotional copy for a product or project: launch and update posts for community platforms and social media, store page descriptions and short blurbs, landing page headlines and calls to action, press-style announcements, and the naming of a product for another language or market. Covers what may be disclosed publicly, keeping claims traceable to shipped changes, and writing a headline that carries information instead of hype. Use when drafting or revising anything aimed at people who do not use the product yet, and when deciding whether a link, key, price, or unreleased detail can appear in public copy.
The English summary is being prepared. Showing the original description.