← Back to Skills Marketplace
metahuan

Yufluentcn Chat Assist

by metahuan · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ Security Clean
34
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install yufluentcn-chat-assist
Description
跨境电商买家消息智能回复助手,支持Amazon、Shopify、TikTok,生成符合买家站内信规范的客服回复草稿。
README (SKILL.md)

跨境客服回复助手

Amazon / Shopify / TikTok Shop 买家站内信/客服消息生成 可直接粘贴发送的回复正文ClawHub / OpenClaw 云端模式 — Harness chat_reply + messaging-guard 在 Yufluent 服务端执行;本机只需 TOKENAPI_KEYtk-*)与 requests

本技能 连接卖家后台、不自动发送消息、不查询真实物流 API;order_context 由你提供摘要,模型不会虚构已核实的状态。


OpenClaw 与 Yufluent(必读)

OpenClaw 对话与技能调用共用同一 tk-*。接入见 https://claw.changzhiai.com/app/openclaw

走哪里 干什么
OpenClaw 对话 Yufluent /v1/chat/completions(同一 tk-*) 收集买家原文与订单上下文、调 run.py、人工微调后发送
客服正式输出 POST /v1/skills/chat-assist/run(同一 tk-*) Harness + 平台规则 → 回复正文

Agent 硬性规则:

  1. 禁止用对话模型自行撰写完整客服回复(尤其是含退款/赔偿承诺的句子)。
  2. 必须通过 python scripts/run.py ...(或 POST /v1/skills/chat-assist/run)获取输出。
  3. 对话模型仅用于:确认平台与语言、收集 message / order_context、提醒卖家审核、建议转人工场景。
  4. 只需 TOKENAPI_KEY不要要求用户另配厂商 LLM Key。

Instructions(Agent 工作流)

  1. 确认平台amazon | shopify | tiktok(决定模板与 messaging 规则)。
  2. 确认回复语言 --lang:默认与买家消息一致;卖家指定时用 zh|en|es|de|fr|ja
  3. 收集输入
    • message(必填):买家原文,原样粘贴,不要改写成第三人称。
    • order_context(强烈建议):订单号、物流单号、当前状态摘要、卖家授权政策(如「可补发」)。
    • product(可选):SKU/品名,便于对齐 Listing。
  4. 调用(必须 — 云端)
    python scripts/run.py \
      --message "Where is my package? Order #123-4567890" \
      --platform amazon \
      --lang en \
      --product "Wireless Earbuds Pro" \
      --order-context "Shipped 2026-05-20, tracking 1Z999, in transit US"
    
  5. 交付:将 stdout 回复正文交给卖家;提醒删除不实承诺、按平台按钮发送。
  6. 计费:402 余额不足;401 密钥无效。

买家消息输入格式

字段 CLI / API 说明
买家消息 --message / message 必填;支持多行;保留买家原语言
订单/物流 --order-context / order_context 建议格式见下表
产品 --product / product 关联 ASIN/SKU 或商品名
平台 --platform amazon | shopify | tiktok,默认 amazon
语言 --lang 输出语言;未指定时 Harness 结合买家消息与 locale

order_context 推荐写法(一行或短段落)

Order #123-4567890 | Shipped 2026-05-20 | Tracking 1Z999AA10123456784 | In transit to US
片段 示例
订单号 Order #123-4567890
物流状态 Delivered / In transit / Delayed — carrier investigation
卖家授权 Seller authorized: replacement once / No refund without return

未提供时 Harness 填「(买家未提供订单号)」—— 回复会以「为您核实」为主,避免编造物流细节。

常见场景(Agent 推断平台与语气)

场景 回复策略要点
物流查询 共情 + 说明已记录单号 + 引导查 tracking / 承诺跟进时效(不承诺具体送达日)
未收到/丢件 请照片/地址确认 + 转核实 + 按授权提及补发/退款流程
质量/损坏 致歉 + 要订单号与照片 + 不预判责任
退款/退货 引用平台 RMA 步骤;无授权时 承诺金额
售前咨询 规格/库存中性说明 + 链到 Listing(Amazon 避免站外营销外链)
差评/威胁 保持专业;要求改评;引导站内解决

输出说明

  • 纯文本回复正文(非 JSON):无标题、无「以下是回复」类前缀。
  • 长度通常 2–5 句(由 messaging-guard 约束)。
  • 人工审核 后再在 Seller Central / Shopify Inbox / TikTok 商家后台发送。

平台差异(简要)

平台 注意
Amazon Buyer-Seller Messaging;避免站外联系与营销外链
Shopify 可提及店铺政策,勿编造条款原文
TikTok Shop 注意 SLA 表述,避免绝对送达日期

与其他技能联动

需求 技能
批量评论洞察后再回信 yufluentcn-review-intel → 本技能
改 Listing 减少售前咨询 yufluentcn-ecommerce-listing
关键词/描述与买家问法对齐 yufluentcn-seo-pro
独立站运营话术策略 yufluentcn-shopify-operator

环境变量

变量 必填 说明
TOKENAPI_KEY tk-*
TOKENAPI_BASE_URL 默认 http://localhost:8080/v1

触发词

  • "回复这条买家消息"
  • "客服怎么回" / "写个英文回复"
  • "订单到哪了" / "refund request"
  • "包裹破损怎么回"
  • "TikTok 买家催发货"

Examples

Amazon — 物流(英文)

python scripts/run.py \
  --message "Where is my order? It's been 10 days." \
  --platform amazon \
  --lang en \
  --order-context "Order #111-2223333, shipped May 18, tracking in transit" \
  --product "USB-C Hub 7-in-1"

Shopify — 退款咨询(中文)

python scripts/run.py \
  --message "想退货,不喜欢颜色" \
  --platform shopify \
  --lang zh \
  --order-context "Order #1042, delivered May 20, 30-day return policy applies" \
  --product "北欧实木餐椅"

TikTok — 售前(英文)

python scripts/run.py \
  --message "Does this fit iPhone 15?" \
  --platform tiktok \
  --lang en \
  --product "MagSafe Phone Case"

合规声明

  • 回复须符合各平台 Buyer-Seller Messaging 与当地消费者法规。
  • 不得索要站外联系方式、引导删改差评、未经授权承诺退款/赔偿。
  • 卖家对发送内容负最终责任;本技能输出仅为草稿。

Harness 合规护栏:harness/scenes/chat_reply/policies/messaging-guard.md


版本记录

版本 日期 变更
v0.3.0 2026-05-28 按 ad-optimize 结构扩展:工作流、输入格式、场景表、示例与合规
v0.2.0 Harness chat_reply 三平台模板
v0.1.0 初始云端薄客户端
Usage Guidance
Install only if you are comfortable sending buyer messages and any order or tracking context you provide to Yufluent's cloud service. Avoid pasting unnecessary personal data, review every generated draft before sending, and consider tightening invocation prompts around ecommerce buyer-message scenarios.
Capability Tags
requires-sensitive-credentials
Capability Assessment
Purpose & Capability
The stated purpose is to generate Amazon, Shopify, and TikTok Shop buyer-message reply drafts; the code sends only user-supplied message, platform, language, product, and optional order context to the Yufluent skill endpoint and prints the returned draft.
Instruction Scope
The workflow tells the agent to confirm platform, language, buyer message, and order context before calling the script, but some trigger phrases such as generic English-reply or refund wording are broad and could benefit from tighter ecommerce-specific scoping.
Install Mechanism
Installation is limited to the requests Python package and a required TOKENAPI_KEY environment variable; no installer scripts, shell persistence, or broad system changes are present.
Credentials
Cloud processing and use of a bearer API token are disclosed in SKILL.md and README and are proportionate to a cloud-based reply-generation service; the skill does not connect to seller accounts, send replies, or query logistics systems itself.
Persistence & Privilege
No background worker, startup persistence, credential harvesting, local indexing, file mutation, or privilege escalation behavior was found. An outcome-reporting helper exists in the shared API module but is not invoked by this skill's runtime path.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install yufluentcn-chat-assist
  3. After installation, invoke the skill by name or use /yufluentcn-chat-assist
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
yufluentcn-chat-assist 1.0.0 - Initial release of a cross-border e-commerce customer service reply assistant for Amazon, Shopify, and TikTok Shop. - Generates compliant, ready-to-send reply drafts based on buyer messages, order context, and platform-specific rules. - Cloud-based processing via Yufluent Harness and messaging-guard; requires only TOKENAPI_KEY and requests on client-side. - Multi-language support (zh, en, es, de, fr, ja) and structured input for message, order details, and product. - Does not connect to seller platforms or fetch real logistics; focuses on draft generation and compliance. - Comprehensive documentation on workflow, input format, compliance, and agent operating rules.
Metadata
Slug yufluentcn-chat-assist
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Yufluentcn Chat Assist?

跨境电商买家消息智能回复助手,支持Amazon、Shopify、TikTok,生成符合买家站内信规范的客服回复草稿。 It is an AI Agent Skill for Claude Code / OpenClaw, with 34 downloads so far.

How do I install Yufluentcn Chat Assist?

Run "/install yufluentcn-chat-assist" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Yufluentcn Chat Assist free?

Yes, Yufluentcn Chat Assist is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Yufluentcn Chat Assist support?

Yufluentcn Chat Assist is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Yufluentcn Chat Assist?

It is built and maintained by metahuan (@metahuan); the current version is v1.0.0.

💬 Comments