抖音文案提取

从抖音链接提取完整文案并生成Word文件。触发条件：用户发送抖音链接并要求提取文案/文本/内容，或说'抓文案''提取内容''做成文件'。支持视频和图文笔记。不做总结不改写，原文案完整提取。

Security Scan

The skill's code and instructions align with its stated purpose (extract Douyin text and produce a .docx); there are no requests for credentials or suspicious network destinations, but the skill omits declaring a required Python dependency and will automatically send the generated file without asking the user.

Purpose & Capability

Name/description (extract Douyin text and produce Word) matches the included instructions and script: curl to fetch the page, extract _ROUTER_DATA JSON, use scripts/create_docx.py to generate a .docx. No unrelated credentials, binaries, or configuration are requested.

Instruction Scope

Instructions stay within the stated task (download Douyin short link HTML, extract JSON, produce and send a Word file). Two caveats: (1) SKILL.md uses a placeholder short-link and does not provide the exact extraction command/regex for _ROUTER_DATA (functional gap, not necessarily malicious); (2) the core rule '先交文件再说话' (always deliver the file first, without asking) is a privacy/UX concern because the agent will send the file automatically.

Install Mechanism

No install spec (instruction-only) and no external downloads — low risk. However, the script depends on python-docx but the skill metadata does not declare this dependency; the script prints a pip install suggestion (including the unusual '--break-system-packages' flag) which is an operational detail users should be aware of.

Credentials

The skill requests no environment variables, credentials, or config paths. The script only reads inputs passed on the command line and writes a .docx to the filesystem (/tmp or specified path). No secrets are requested or used.

Persistence & Privilege

always is false and the skill does not request persistent or elevated privileges nor modify other skills or system-wide configs. It writes only the output .docx to a specified path and sends it via the agent's message action.

Assessment

This skill appears to do what it says: fetch a Douyin short link page, extract the _ROUTER_DATA JSON, convert the extracted text into a .docx, and send the file. Before installing, note: (1) the Python script requires the python-docx package (metadata does not declare this); install python-docx in a safe environment rather than using '--break-system-packages'. (2) The skill will deliver the generated file automatically ('先交文件再说话') — if that surprises you, ask for a modification so it prompts for user confirmation before sending. (3) The SKILL.md omits the exact extraction steps for _ROUTER_DATA and may fail if Douyin requires JS rendering; this is a functional limitation, not a security issue. (4) Ensure you trust the agent context where files are sent (avoid sending sensitive links/content if you don't want automatic upload).

抖音文案提取技能

触发条件

用户发送抖音链接（https://v.douyin.com/xxx 或分享口令）+ 提取/抓取/文案/内容/文件 相关意图。

工作流程

Step 1: 解析短链获取数据

curl -s -L -o /tmp/douyin_page.html \
  -w "%{http_code}" \
  "https://v.douyin.com/xxx/" \
  -H "User-Agent: Mozilla/5.0 (iPhone; CPU iPhone OS 16_0 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/16.0 Mobile/15E148 Safari/604.1" \
  --connect-timeout 10

Step 2: 提取 _ROUTER_DATA

从 HTML 中提取 _ROUTER_DATA JSON，关键字段：

• item_list[0].desc — 完整文案（核心，一个字不改）
• item_list[0].author.nickname — 作者昵称
• item_list[0].statistics — 点赞/评论/转发/收藏
• item_list[0].create_time — 发布时间戳

Step 3: 生成 Word 文件

使用 scripts/create_docx.py：

python3 scripts/create_docx.py --title "标题" --author "作者" --content "完整文案" --output /tmp/output.docx

Step 4: 投递文件

message action=send filePath=/tmp/output.docx

核心规则

1. 原文案完整提取，不总结不改写
2. 先交文件再说话 — 不问"要不要做"
3. 按原文结构排版：识别"第X组""复制这句"等结构化内容
4. 文件名包含作者名便于识别