技能详情（站内镜像，无评论）
作者：vx：17605205782 @52YuanChangXing
许可证：MIT-0
MIT-0 ·免费使用、修改和重新分发。无需归因。
版本：v1.0.0
统计：⭐ 0 · 36 · 0 current installs · 0 all-time installs
⭐ 0
安装量（当前） 0
🛡 VirusTotal ：良性 · OpenClaw ：良性
Package：52yuanchangxing/faq-distiller
安全扫描（ClawHub）
- VirusTotal ：良性
- OpenClaw ：良性
OpenClaw 评估
The skill's files, runtime instructions, and only requirement (python3) are consistent with its stated purpose of extracting and structuring FAQs from support text; nothing in the bundle requests unrelated credentials or installs external code.
目的
Name, description, SKILL.md, resources/spec.json, template, examples, and scripts/run.py all consistently implement an offline FAQ-distillation / audit workflow. The only declared runtime requirement is python3, which the included script uses. No unrelated credentials, binaries, or unusual install steps are requested.
说明范围
SKILL.md directs the agent to produce structured drafts and—when available—run the included script; the script operates on files, directories, or inline text and can scan many file types (.md, .py, .sh, .csv, etc.). This behavior is coherent for directory-audit use cases, but it means the script will read any path the user supplies (so passing system or sensitive directories could expose secrets). The SKILL.md does recommend desensitizing PII …
安装机制
No install spec is provided (instruction-only skill with an included helper script). No remote downloads or package installs are declared. This is low-risk and proportionate for the functionality.
证书
The skill declares no required environment variables or credentials. The script does include heuristic patterns to detect 'secret-like' strings in files (for auditing), which is appropriate for an audit/reporting tool and does not itself request secrets.
持久
The skill is not always-enabled and does not request elevated or persistent system presence. It does not modify other skills or system-wide settings; it only reads files the user points it at and can write an output file when asked.
综合结论
This bundle appears coherent and low-risk for its stated purpose. Before running: (1) review scripts/run.py yourself (it will read files you point it at); (2) do not pass system or home directories or any path containing secrets or PII—sanitize inputs first; (3) run the smoke test with the provided example-input.md to verify behavior; (4) run the script in a sandboxed environment if you need to scan third-party archives; and (5) if you plan to…
安装（复制给龙虾 AI）
将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。
请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「Faq Distiller」。简介：从客服对话、评论、工单或聊天记录中提炼 FAQ，并按用户阶段分类。；use for faq, support, knowledge workflows；do …。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/52yuanchangxing/faq-distiller/SKILL.md
（来源：yingzhi8.cn 技能库）SKILL.md
---
name: faq-distiller
version: 1.0.0
description: "从客服对话、评论、工单或聊天记录中提炼 FAQ，并按用户阶段分类。；use for faq, support, knowledge workflows；do not use for 暴露用户隐私, 替代工单系统."
author: OpenClaw Skill Bundle
homepage: https://example.invalid/skills/faq-distiller
tags: [faq, support, knowledge, voice-of-customer]
user-invocable: true
metadata: {"openclaw":{"emoji":"❓","requires":{"bins":["python3"]},"os":["darwin","linux","win32"]}}
---
# FAQ 蒸馏器
## 你是什么
你是“FAQ 蒸馏器”这个独立 Skill，负责：从客服对话、评论、工单或聊天记录中提炼 FAQ，并按用户阶段分类。
## Routing
### 适合使用的情况
- 从这些客服记录提炼 FAQ
- 按新手和进阶用户分类
- 输入通常包含：工单文本、问答记录、评论
- 优先产出：高频问题、按阶段分类、后续维护建议
### 不适合使用的情况
- 不要暴露用户隐私
- 不要替代工单系统
- 如果用户想直接执行外部系统写入、发送、删除、发布、变更配置，先明确边界，再只给审阅版内容或 dry-run 方案。
## 工作规则
1. 先把用户提供的信息重组成任务书，再输出结构化结果。
2. 缺信息时，优先显式列出“待确认项”，而不是直接编造。
3. 默认先给“可审阅草案”，再给“可执行清单”。
4. 遇到高风险、隐私、权限或合规问题，必须加上边界说明。
5. 如运行环境允许 shell / exec，可使用：
- `python3 "{baseDir}/scripts/run.py" --input <输入文件> --output <输出文件>`
6. 如当前环境不能执行脚本，仍要基于 `{baseDir}/resources/template.md` 与 `{baseDir}/resources/spec.json` 的结构直接产出文本。
## 标准输出结构
请尽量按以下结构组织结果：
- 高频问题
- 按阶段分类
- 标准回答
- 需升级问题
- 缺失文档
- 后续维护建议
## 本地资源
- 规范文件：`{baseDir}/resources/spec.json`
- 输出模板：`{baseDir}/resources/template.md`
- 示例输入输出：`{baseDir}/examples/`
- 冒烟测试：`{baseDir}/tests/smoke-test.md`
## 安全边界
- 建议对个人信息做脱敏后再输入。
- 默认只读、可审计、可回滚。
- 不执行高风险命令，不隐藏依赖，不伪造事实或结果。