ClawHub

Reply Styles

v1.0.0

Draft high-EQ, style-controlled replies for chat, email, DMs, customer support, community operations, sales, and difficult conversations. Use when the user w...

⭐ 0· 172·0 current·0 all-time
by@tobewin

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for tobewin/reply-styles.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Reply Styles" (tobewin/reply-styles) from ClawHub.
Skill page: https://clawhub.ai/tobewin/reply-styles
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install reply-styles

ClawHub CLI

Package manager switcher

npx clawhub@latest install reply-styles
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
medium confidence
ℹ
Purpose & Capability
The name/description (style-driven reply drafting) match the included content: style catalogs, examples, and scripts that assemble reply bundles. One minor mismatch: the runtime instructions and provided scripts assume a Node.js runtime (commands like `node scripts/*.mjs`), but the skill's registry metadata lists no required binaries. Otherwise the declared requirements (no env vars, no credentials) are proportional to the stated purpose.
✓
Instruction Scope
SKILL.md confines agent actions to reading the bundled reference files and running the included Node scripts to build reply bundles. There are no instructions to read arbitrary system files, access unrelated environment variables, or transmit data to external endpoints.
✓
Install Mechanism
No install spec is provided (instruction-only plus local scripts), so nothing is downloaded or written to disk beyond the skill bundle. The included scripts are local JavaScript modules; no external installers or remote archives are used.
✓
Credentials
The skill requests no environment variables or credentials and the scripts do not reference secrets. The lack of any declared credentials is appropriate for a purely stylistic reply generator.
✓
Persistence & Privilege
The skill is not always-enabled and does not request elevated persistence or modify other skills. Default autonomous invocation is allowed (platform default) and is consistent with the skill's utility.
Assessment
This skill appears to be a local, data-driven reply formatter and is internally consistent with that purpose. Before installing: ensure the agent environment provides Node.js if you want to run the provided scripts (SKILL.md examples use `node scripts/*.mjs`), since the skill metadata doesn't declare Node as a required binary. The bundle contains only reference docs and local JS modules — no credentials, no network endpoints, and no install downloader — so it does not appear to exfiltrate data. As always, review example outputs before sending them to others and consider whether you want the agent to use the skill autonomously (the default platform behavior). If you need absolute certainty, ask the author to add Node to the required-binaries list or provide an explicit note about runtime requirements.

Like a lobster shell, security has layers — review code before you run it.

Runtime requirements

💬 Clawdis
latestvk97d7emttghystktyanw8h67fx83bhgn
172downloads
0stars
1versions
Updated 1mo ago
v1.0.0
MIT-0
@tobewin
latest
Download

Reply Styles

为用户生成高情商、风格化、可直接发送的回复文案。

核心定位：

  • 模型无关
  • 预置风格，不靠用户反复调教
  • 同时支持高情商和明确边界
  • 可用于私聊、邮件、客服、销售、社群、管理沟通

何时使用

  • 用户说“帮我高情商回复一下”
  • 用户要“委婉一点、坚定一点、像高管一点、像客服一点”
  • 用户要回复客户、老板、同事、朋友、社群成员
  • 用户要拒绝、安抚、催促、解释、道歉、成交、收口

工作方式

主 SKILL.md 只保留路由与规则。按需读取：

  • 风格总览：references/style-catalog.md
  • 场景映射：references/scenario-map.md
  • 强度系统：references/intensity-system.md
  • 渠道映射：references/channel-map.md
  • 人格包：references/persona-packs.md
  • 回复模式：references/mode-packs.md
  • 通用原则：references/core-rules.md
  • 风格示例：references/style-examples.md
  • 效果验收：references/effect-rubric.md

如果要快速定位风格或场景，优先运行：

  • node scripts/get-style-bundle.mjs --list
  • node scripts/get-style-bundle.mjs --style warm-professional
  • node scripts/get-style-bundle.mjs --list-personas
  • node scripts/get-style-bundle.mjs --list-modes
  • node scripts/get-style-bundle.mjs --style warm-professional --intensity balanced --channel wechat
  • node scripts/recommend-style.mjs --scene angry-customer --goal deescalate --channel support-ticket --persona calm-operator --mode soothe-first
  • node scripts/scaffold-reply.mjs --style firm-boundary --scene reject-request --intensity soft --channel wechat --persona trusted-advisor --mode clear-reject
  • node scripts/showcase-replies.mjs --scene reject-request --intent "礼貌拒绝继续无偿帮忙" --persona trusted-advisor --mode clear-reject

生成原则

  1. 先判断关系，再判断情绪温度，再判断风格、强度和渠道。
  2. 高情商不等于软弱，不要为了“礼貌”牺牲边界。
  3. 先保留用户真实意图，再做语气优化。
  4. 默认输出一版可以直接发的正文；只有用户要求时再给多个版本。
  5. 不要解释“我为什么这样写”，除非用户明确要分析。
  6. 风格要稳定，不要一条回复里又像客服又像朋友。

最小决策流程

Step 1：识别场景

从用户输入中提取：

  • 对象：客户 / 老板 / 同事 / 朋友 / 社群成员 / 陌生人
  • 目标：安抚 / 拒绝 / 催促 / 成交 / 道歉 / 解释 / 推进 / 收口
  • 风险：对方生气 / 关系脆弱 / 场合正式 / 需要边界 / 需要成交
  • 调性：温暖 / 专业 / 坚定 / 高级 / 轻松 / 干练

Step 2：选风格

优先先跑脚本拿最小推荐结果，再读 references/style-catalog.md。

如果用户明确指定风格，再直接读取对应风格说明：

  • warm-professional
  • high-eq-soft
  • firm-boundary
  • de-escalation
  • executive-crisp
  • founder-direct
  • luxury-concierge
  • community-manager
  • premium-sales
  • supportive-coach
  • playful-friendly
  • relationship-repair
  • consultant-polished
  • partner-diplomatic
  • pr-safe
  • recruiter-polite
  • operations-push
  • customer-success
  • elder-respectful
  • cool-minimal
  • bookish-gentle
  • service-recovery
  • brand-host
  • legal-clean

Step 3：选强度和渠道

优先读：

  • references/intensity-system.md
  • references/channel-map.md
  • references/persona-packs.md
  • references/mode-packs.md

默认：

  • 强度：balanced
  • 渠道：wechat
  • 人格包：留空
  • 回复模式：留空

Step 4：按结构写回复

默认结构：

  1. 开场先接住情绪或意图
  2. 中段表达核心信息
  3. 结尾给下一步或收口

Step 5：对照原则自检

读 references/core-rules.md 和 references/effect-rubric.md，确保：

  • 不油腻
  • 不虚伪
  • 不啰嗦
  • 不失边界
  • 适合实际发送
  • 风格稳定
  • 像真人会发的话

输出偏好

默认输出：

  • 1 个成稿版本
  • 直接可发送
  • 不带多余解释
  • 保持用户原意

用户如果要求，可额外提供：

  • 更软一版
  • 更硬一版
  • 更短一版
  • 更高级一版
  • 邮件版 / 微信版 / 评论区版

不要这样做

  • 不要一味加“哈哈”“呀”“呢”制造假高情商
  • 不要把简单回复写成鸡汤
  • 不要过度道歉
  • 不要把坚定边界写成攻击
  • 不要把销售感写成油腻推销
  • 不要让风格压过信息本身

成功标准

这项 skill 的目标不是“变得更礼貌”，而是让用户觉得：

  • 这句话像我想说的，只是更高级
  • 既体面又有效
  • 有风格，但不做作
  • 不需要再自己手动调语气

当前系统支持：

  • 24 套主风格
  • 6 个人格包
  • 6 个回复模式
  • 3 档强度：soft / balanced / sharp
  • 6 个渠道：wechat / email / dm / group / comment / support-ticket
  • 共 24 × 6 × 6 × 3 × 6 = 15552 种组合方式

Comments

Loading comments...