Ai customer support
客服回复助手适合运营、产品、销售、software在用户提出“客服怎么回更稳”这类问题,需要快速拆解目标、判断重点并形成可执行结果时使用,帮助基于输入材料生成销售策略、沟通素材、跟进计划。From its SKILL.md
npx -y skills add skillsaiagent/aiskills --skill ai-customer-supportAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
SKILL.md
4.9 KB, ~1.6k tokens by cl100k_base, as published. Nobody here has run it
ai-customer-support 客服回复助手
概述
客服回复助手用于回答「客服怎么回更稳」、资料整理、数据清理、行动项,适合运营、产品、销售、software在明确业务目标、内容材料或分析对象后调用。 它会结合文档、表格或记录、粘贴文档、表格、会议记录、客户反馈、评论文本或待整理资料。等输入,整理关键上下文,并输出销售策略、沟通素材、跟进计划,便于继续执行、复盘或交付。
什么时候使用
适用场景
- 用户提出“客服怎么回更稳”这类问题,需要快速拆解目标、判断重点并形成可执行结果
- 运营、产品、销售、software需要围绕客服回复助手生成销售策略、沟通素材、跟进计划
- 用户已经准备了整理目标(说明希望得到的结构摘要、清理规则、行动项、缺口提醒或交付用途。)、使用对象(说明整理结果要服务的人群、团队、角色或业务场景。)、资料链接(填写公开可访问的文档、表格、页面或资料链接;受限内容请改为上传或粘贴。),希望整理成可执行的分析或优化结果
- 用户需要把客服回复助手相关材料转成清晰结论、优先级和下一步动作
调用方式
通过导出的 Python runner 直接调用 AI Skills API:
命令示例
基础调用
python3 scripts/run.py --params '{}'
带常用参数调用
python3 scripts/run.py --params '{"goal":"整理目标"}'
参数说明
| 参数 | 类型 | 必填 | 默认 | 说明 |
|---|---|---|---|---|
goal | string | 否 | - | 说明希望得到的结构摘要、清理规则、行动项、缺口提醒或交付用途 |
audience | string | 否 | - | 说明整理结果要服务的人群、团队、角色或业务场景 |
materialUrl | string | 否 | - | 填写公开可访问的文档、表格、页面或资料链接;受限内容请改为上传或粘贴;需要传可访问的完整 URL |
materialFile | string | 否 | - | 上传文档、表格、会议纪要、客户资料或评论数据文件 |
materialText | string | 否 | - | 粘贴文档、表格、会议记录、客户反馈、评论文本或待整理资料 |
brandRequirements | string | 否 | - | 补充字段口径、时间范围、格式要求、不可改动事实、敏感信息处理或人工复核重点 |
完整机器可读参数结构见 references/form-schema.json。
参数取值参考
当前技能没有需要额外查表的分类参数。
支持的输入格式
当前技能直接接收 JSON 参数;如果参数里包含链接字段,请传完整、可访问的 URL。
示例请求
下面的示例参数可直接传给 scripts/run.py,runner 会把它们发送给 AI Skills API。
python3 scripts/run.py --params '{"goal":"整理目标"}'
等价的 --params JSON:
{
"goal": "整理目标"
}
返回结果示例
{
"success": true,
"data": {
"message": "示例结果请以技能真实返回结构为准。"
},
"meta": {
"executionTime": 842,
"cached": false
}
}
交付内容
- 销售策略、沟通素材、跟进计划:围绕用户目标整理可直接阅读、复盘或交付的核心结果。
- 输入材料解读:结合整理目标(说明希望得到的结构摘要、清理规则、行动项、缺口提醒或交付用途。)、使用对象(说明整理结果要服务的人群、团队、角色或业务场景。)、资料链接(填写公开可访问的文档、表格、页面或资料链接;受限内容请改为上传或粘贴。)提炼关键上下文和判断依据。
- 下一步动作:给出优先级、执行建议或可继续加工的内容框架。
结果使用建议
- 先判断输出是否回答了用户关于「客服回复助手」的核心问题。
- 再检查结果是否覆盖销售策略、沟通素材、跟进计划,以及是否给出明确下一步动作。
- 如果输入材料较少,建议让用户补充目标、受众、限制条件或原始材料后再运行。
运行前准备
AISKILLS_BASE_URL:默认https://ai-skills.aiAISKILLS_API_KEY:必填,用于认证调用AISKILLS_TENANT_ID:默认default
What ships with it: 4 files
15.5 KB alongside SKILL.md, 1 of them executable
agents/
- openai.yaml210 B
references/
- form-schema.json2.6 KB
- skill.json7.8 KB
scripts/
- run.pyruns4.9 KB
Gives 0 of the 12 instructions most customer support skills give in ~1.6k tokens
Counted across 123 of the 124 authors here whose files we hold, read 2026-08-07
- Call RUBE_SEARCH_TOOLS first to get current schemasin 12 of 123, across 4 files
- Confirm connection status is ACTIVE before running workflowsin 12 of 123, across 4 files
- Stop and ask for clarification if required inputs are missingin 9 of 123, across 2 files
- Call RUBE_MANAGE_CONNECTIONS with the helpdesk toolkitin 9 of 123, across 2 files
- Use both timestamp and ID for cursor navigationin 8 of 123, across 1 file
- Implement backoff on 429 responsesin 8 of 123, across 1 file
- Parse response data defensively with fallback patternsin 8 of 123, across 1 file
- Use this skill only when the task clearly matches the scopein 8 of 123, across 1 file
- Pass a JSON file as the positional argumentin 7 of 123, across 1 file
- Specify output format with the --format flagin 7 of 123, across 1 file
- Run health, churn, and expansion scripts togetherin 7 of 123, across 1 file
- Verify output files contain expected records before continuingin 7 of 123, across 1 file
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.