Back to Details

douyin-prohibited-word

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

抖音违禁词检测

Douyin Prohibited Word Detection

简介

Overview

抖音违禁词检测是一款专为抖音内容创作者和运营人员设计的智能合规检测工具，帮你快速排查文案中的违禁词并给出安全替换方案。
视频被限流、文案被下架？可能就是一两个词惹的祸。把文案扔进来，快速告诉你哪里踩雷、怎么改才安全。
通过简单的输入操作，你可以：
  • 🔍 自动识别抖音违禁词，原文加粗定位风险点
  • 💡 获得结合上下文的智能替换建议 + 更换理由
  • ✏️ 一键生成替换后的可发布版本，复制粘贴就能用
适用于短视频编导、电商运营、直播策划、品牌市场等需要确保抖音文案合规发布的场景。
Douyin Prohibited Word Detection is an intelligent compliance detection tool designed specifically for Douyin content creators and operators, helping you quickly identify prohibited words in copy and provide safe replacement solutions.
Video restricted? Copy taken down? It might be caused by just one or two words. Paste your copy here to quickly find out where you've violated rules and how to revise it safely.
With simple input operations, you can:
  • 🔍 Automatically identify Douyin prohibited words and locate risk points by boldening them in the original text
  • 💡 Obtain context-aware intelligent replacement suggestions + reasons for replacement
  • ✏️ Generate a publishable version with replacements in one click, ready to copy and use
It applies to scenarios where compliance of Douyin copy needs to be ensured, such as short video directors, e-commerce operators, live stream planners, brand marketing teams, etc.

功能特性

Features

🎯 核心功能

🎯 Core Functions

  • 🔍 违禁词标记：原文中命中词直接加粗，一眼看到风险点
  • 💡 智能替换建议：每个违禁词配一个结合上下文的替换词 + 更换理由，帮你"改通顺"而不是机械替换
  • ✏️ 优化文案生成：直接输出一份替换后的可发布版本，替换词加粗标记，复制粘贴就能用
  • 📎 文件自动交付：自动生成纯文本优化文案文件，点击即可下载保存
  • 📏 长文案分批检测：超过 3000 字自动提醒，支持分批检测后自动合并结果
  • 🔍 Prohibited Word Marking: Hit words in the original text are directly boldened, allowing you to spot risk points at a glance
  • 💡 Intelligent Replacement Suggestions: Each prohibited word is paired with a context-aware replacement word + reason for replacement, helping you "revise smoothly" instead of mechanical replacement
  • ✏️ Optimized Copy Generation: Directly output a publishable version after replacement, with replacement words boldened for easy identification, ready to copy and use
  • 📎 Automatic File Delivery: Automatically generate a plain text optimized copy file, which can be downloaded and saved with a click
  • 📏 Batch Detection for Long Copy: Automatically reminds when the copy exceeds 3000 words, supports batch detection and automatic result merging

✨ 特色亮点

✨ Highlight Features

  • 📥 多输入形态：支持直接粘贴文案、上传 TXT/DOC/DOCX 文件、上传图片提取文字、粘贴网页链接
  • ⚡ 一站式处理：文本提取 + 违禁词检测一次完成，无需多次调用脚本
  • 🌐 动态网页支持：网页检测使用 Playwright 无头浏览器渲染 JS 页面，也支持静态页面回退
  • 🔤 英文误匹配过滤：内置英文单词误匹配过滤，不会把正常英文单词片段误判为违禁词
  • 🔄 网络自动重试：遇到网络波动自动重试最多 3 次，无需手动干预
  • 📥 Multiple Input Forms: Supports direct copy pasting, uploading TXT/DOC/DOCX files, uploading images to extract text, and pasting web links
  • ⚡ One-stop Processing: Text extraction + prohibited word detection completed in one go, no need to call scripts multiple times
  • 🌐 Dynamic Webpage Support: Webpage detection uses Playwright headless browser to render JS pages, also supports fallback to static pages
  • 🔤 English False Match Filtering: Built-in filtering for English word false matches, won't misjudge normal English word fragments as prohibited words
  • 🔄 Automatic Network Retry: Automatically retries up to 3 times when encountering network fluctuations, no manual intervention required

一键安装

One-click Installation

前置条件

Prerequisites

  • Python 3.8+
  • 已注册 红狐Hub 账号并获取 API Key（格式 
    ak_xxxxxxxx
    ）
  • Python 3.8+
  • Registered RedFox Hub account and obtained API Key (format 
    ak_xxxxxxxx
    )

安装步骤

Installation Steps

bash
undefined
bash
undefined

1. 安装 Python 依赖

1. Install Python dependencies

pip install python-docx==1.1.0 beautifulsoup4==4.12.3 playwright==1.58.0
pip install python-docx==1.1.0 beautifulsoup4==4.12.3 playwright==1.58.0

2. 安装 Playwright 浏览器（网页检测需要）

2. Install Playwright browser (required for webpage detection)

playwright install chromium

---
playwright install chromium

---

鉴权

Authentication

获取 API Key

Obtain API Key

  1. 访问 红狐Hub 官网 了解服务详情
  2. 前往 注册页面 注册账号
  3. 新注册用户将获赠免费积分，可立即开始使用 API 服务
  4. 注册登录后，在个人中心获取 API Key，格式为 
    ak_xxxxxxxx
  1. Visit RedFox Hub Official Website to learn about service details
  2. Go to Registration Page to register an account
  3. New registered users will receive free credits and can start using the API service immediately
  4. After registering and logging in, obtain the API Key in the personal center, formatted as 
    ak_xxxxxxxx

配置 API Key

Configure API Key

将 API Key 配置为环境变量 
REDFOX_API_KEY
：
方式一：临时设置（当前终端会话有效）
bash
export REDFOX_API_KEY=ak_xxxxxxxx
方式二：永久设置（推荐）
bash
undefined
Configure the API Key as the environment variable 
REDFOX_API_KEY
:
Method 1: Temporary Setting (valid for current terminal session)
bash
export REDFOX_API_KEY=ak_xxxxxxxx
Method 2: Permanent Setting (recommended)
bash
undefined

Bash 用户

Bash users

echo 'export REDFOX_API_KEY=ak_xxxxxxxx' >> ~/.bashrc source ~/.bashrc
echo 'export REDFOX_API_KEY=ak_xxxxxxxx' >> ~/.bashrc source ~/.bashrc

Zsh 用户

Zsh users

echo 'export REDFOX_API_KEY=ak_xxxxxxxx' >> ~/.zshrc source ~/.zshrc

| 变量名 | 必填 | 说明 |
|--------|------|------|
| `REDFOX_API_KEY` | 是 | RedFox API 访问密钥，格式 `ak_xxxxxxxx`，脚本自动通过 `X-API-KEY` 请求头附加 |

---
echo 'export REDFOX_API_KEY=ak_xxxxxxxx' >> ~/.zshrc source ~/.zshrc

| Variable Name | Required | Description |
|--------|------|------|
| `REDFOX_API_KEY` | Yes | RedFox API access key, formatted as `ak_xxxxxxxx`, automatically attached via `X-API-KEY` request header by the script |

---

使用指南

Usage Guide

基础使用

Basic Usage

1. 直接贴文案

1. Directly Paste Copy

最简单的用法，直接把抖音文案粘贴进来：
用户：帮我看下这段抖音脚本有没有违禁词：这款美白神器真的太有效了，用了三天就白了一个度
检测完成后自动输出三个板块：检测结果 → 修改建议 → 优化文案，并自动生成可下载的纯文本文件。
The simplest way is to directly paste your Douyin copy:
User: Help me check if there are any prohibited words in this Douyin script: This whitening miracle is really effective, it lightened my skin tone by one shade in just three days
After detection, it will automatically output three sections: Detection Results → Revision Suggestions → Optimized Copy, and automatically generate a downloadable plain text file.

2. 上传文件

2. Upload Files

支持 TXT、DOC、DOCX 文件，直接上传即可自动提取文字并检测：
用户：[上传 script.docx] 帮我检查这个文档
脚本自动提取文件内容 → 字数判断 → 调用检测 API → 输出结果。
Supports TXT, DOC, DOCX files, just upload them to automatically extract text and detect:
User: [Upload script.docx] Help me check this document
The script automatically extracts file content → checks word count → calls detection API → outputs results.

3. 上传图片

3. Upload Images

上传视频截图或字幕截图，自动提取图中文字进行检测（仅提取文字，不分析图片内容）：
用户：[上传字幕截图.png] 检测图片里的文字有没有违禁词
Upload video screenshots or subtitle screenshots, automatically extract text from the image for detection (only extracts text, does not analyze image content):
User: [Upload subtitle screenshot.png] Detect if there are any prohibited words in the text of this image

4. 粘贴链接

4. Paste Links

直接粘贴网页地址，自动抓取页面文字内容进行检测：
用户：帮我检测这个网页 https://example.com/article
Directly paste the webpage address, automatically crawl the page text content for detection:
User: Help me detect this webpage https://example.com/article

高级使用：超长文案分批检测

Advanced Usage: Batch Detection for Extra-long Copy

当文案超过 3000 字时，系统会自动暂停并询问：
  • 回复 1：仅检测前 3000 字符
  • 回复 2：分批检测全部内容，完成后自动合并结果
  • 回复 3：取消
当文案超过 10000 字时，系统会提示建议手动分批，不执行检测。
When the copy exceeds 3000 words, the system will automatically pause and ask:
  • Reply 1: Only detect the first 3000 characters
  • Reply 2: Detect all content in batches and automatically merge results after completion
  • Reply 3: Cancel
When the copy exceeds 10000 words, the system will suggest manual batch processing and will not perform detection.

命令速查

Command Quick Reference

输入形态助手动作
纯文本计算字数 → 字数闸门 → 
--content="..."
→ 三板块输出 → 写文件
上传 TXT/DOC/DOCX
--file=path --extract-only
查字数 → 按规则检测 → 三板块输出 → 写文件
上传图片仅提取图中文字 → 计算字数 → 
--content="提取文字"
→ 三板块输出 → 写文件
粘贴 
https://...
--url=... --extract-only
查字数 → 按规则检测 → 三板块输出 → 写文件
超长文案（3000~10000）发出询问语，收到 1/2/3 后再继续
超长文案（>10000）直接提示建议手动分批，不执行检测
--content
、
--file
、
--url
三者互斥，每次调用仅传一个。平台已固定为抖音，无需指定。
Input FormAssistant Action
Plain TextCalculate word count → word count threshold → 
--content="..."
→ three-section output → write file
Upload TXT/DOC/DOCX
--file=path --extract-only
to check word count → detect according to rules → three-section output → write file
Upload ImageOnly extract text from image → calculate word count → 
--content="extracted text"
→ three-section output → write file
Paste 
https://...
--url=... --extract-only
to check word count → detect according to rules → three-section output → write file
Extra-long Copy (3000~10000 words)Send inquiry, continue only after receiving 1/2/3
Extra-long Copy (>10000 words)Directly suggest manual batch processing and do not perform detection
--content
, 
--file
, 
--url
are mutually exclusive, only one can be passed per call. The platform is fixed as Douyin, no need to specify.

输出格式说明

Output Format Description

检测成功时，输出固定包含三个板块（顺序固定，禁止增减）：
板块内容
🔍 违禁词检测结果检测平台、违禁词数量、违禁词类型
💡 修改建议违禁词 → 替换词 → 更换理由 表格
✏️ 建议优化文案替换后的完整文案，替换处加粗显示
未检测到违禁词时，仅输出「未检测到违禁词，文案内容合规✅」，不输出后两个板块。
重要：三板块输出完成后，会自动将纯文本优化文案写入 
./抖音_优化文案_{随机6位数字}.txt
并以文件卡片形式发送给用户，可直接点击下载。
When detection is successful, the output will include three fixed sections (fixed order, no addition or reduction):
SectionContent
🔍 Prohibited Word Detection ResultsDetection platform, number of prohibited words, types of prohibited words
💡 Revision SuggestionsTable of prohibited word → replacement word → reason for replacement
✏️ Suggested Optimized CopyComplete copy after replacement, with replacement parts boldened
If no prohibited words are detected, only output "No prohibited words detected, copy content is compliant✅", and do not output the latter two sections.
Important: After outputting the three sections, the plain text optimized copy will be automatically written to 
./Douyin_Optimized_Copy_{random 6-digit number}.txt
and sent to the user as a file card, which can be directly clicked to download.

使用场景

Usage Scenarios

场景一：短视频脚本发布前自查

Scenario 1: Self-check before publishing short video scripts

角色：短视频编导/创作者
需求：口播脚本、字幕文案发布前排查极限词、禁宣用语
使用方式：
  1. 将完成的脚本文案粘贴进来
  2. 查看违禁词标记和替换建议
  3. 下载优化后的安全版本
预期收益：避免因违禁词导致视频限流或下架，安心发布
Role: Short video director/creator
Requirement: Check limit words and prohibited promotional terms in voiceover scripts and subtitle copy before publishing
Usage Method:
  1. Paste the completed script copy
  2. View prohibited word markings and replacement suggestions
  3. Download the optimized safe version
Expected Benefit: Avoid video restriction or takedown due to prohibited words, publish with confidence

场景二：电商商品文案批量扫雷

Scenario 2: Batch screening of e-commerce product copy

角色：电商运营
需求：商品详情页、活动页文案上架前批量合规检测
使用方式：
  1. 上传包含多条商品文案的 DOCX 文档
  2. 系统自动分批检测长文档
  3. 获取全部替换后的合规版本
预期收益：批量排查，避免商品因文案违规被下架处罚
Role: E-commerce operator
Requirement: Batch compliance detection of product detail page and activity page copy before going online
Usage Method:
  1. Upload a DOCX document containing multiple product copies
  2. The system automatically detects long documents in batches
  3. Obtain all replaced compliant versions
Expected Benefit: Batch screening, avoid product takedown penalties due to copy violations

场景三：直播话术合规过筛

Scenario 3: Compliance screening of live stream scripts

角色：直播话术策划
需求：直播间台词、促销话术上线前快速过筛
使用方式：
  1. 直接粘贴直播话术脚本
  2. 一键获取违禁词位置和替换建议
  3. 根据建议调整话术
预期收益：保障直播话术合规，降低直播风险
Role: Live stream script planner
Requirement: Quickly screen live stream lines and promotional scripts before going online
Usage Method:
  1. Directly paste the live stream script
  2. Get prohibited word positions and replacement suggestions in one click
  3. Adjust the script according to suggestions
Expected Benefit: Ensure live stream script compliance, reduce live stream risks

场景四：品牌落地页文案抽查

Scenario 4: Spot check of brand landing page copy

角色：品牌/市场团队
需求：落地页、H5 活动页发布前文案合规抽查
使用方式：
  1. 粘贴落地页 URL
  2. 系统自动抓取页面文字内容
  3. 检测文案合规性
预期收益：快速完成页面文案审核，避免对外发布后出现问题
Role: Brand/marketing team
Requirement: Compliance spot check of landing page and H5 activity page copy before publishing
Usage Method:
  1. Paste the landing page URL
  2. The system automatically crawls the page text content
  3. Detect copy compliance
Expected Benefit: Complete page copy review quickly, avoid problems after public release

项目架构

Project Architecture

目录结构

Directory Structure

douyin-prohibited-word/
├── SKILL.md                             # Skill 说明文档（本文件）
├── references/
│   └── core_workflow.md                 # 完整执行规程（输出模板、铁律、详细步骤）
├── scripts/
│   └── check_sensitive_words.py         # 核心检测脚本
└── _meta.json                           # 元数据配置
douyin-prohibited-word/
├── SKILL.md                             # Skill documentation (this file)
├── references/
│   └── core_workflow.md                 # Complete execution procedures (output templates, rules, detailed steps)
├── scripts/
│   └── check_sensitive_words.py         # Core detection script
└── _meta.json                           # Metadata configuration

技术栈

Technology Stack

组件技术
运行环境Python 3.8+
HTTP 请求requests（Python 原生 HTTP 库，自动重试）
文档解析python-docx（DOC/DOCX）
网页提取Playwright（JS 动态页面）+ BeautifulSoup4（静态页面回退）
检测平台抖音（硬编码）
ComponentTechnology
Runtime EnvironmentPython 3.8+
HTTP Requestrequests (Python native HTTP library, automatic retry)
Document Parsingpython-docx (DOC/DOCX)
Webpage ExtractionPlaywright (JS dynamic pages) + BeautifulSoup4 (static page fallback)
Detection PlatformDouyin (hard-coded)

核心模块说明

Core Module Description

模块文件职责
文本提取
check_sensitive_words.py
— 
extract_from_file()
/ 
extract_from_web()
从文件、网页中提取文本内容
违禁词检测
check_sensitive_words.py
— 
check_sensitive_words()
调用违禁词检测 API，返回标记后的 HTML + 违禁词列表
英文误匹配过滤
check_sensitive_words.py
自动过滤英文单词内部子串误标（如 "Glasswing" 中的 "ass"）
执行规程
references/core_workflow.md
完整的输出模板、执行铁律、超长内容处理规则、异常处理话术
ModuleFileResponsibility
Text Extraction
check_sensitive_words.py
— 
extract_from_file()
/ 
extract_from_web()
Extract text content from files and webpages
Prohibited Word Detection
check_sensitive_words.py
— 
check_sensitive_words()
Call prohibited word detection API, return marked HTML + prohibited word list
English False Match Filtering
check_sensitive_words.py
Automatically filter false marking of substrings inside English words (e.g., "ass" in "Glasswing")
Execution Procedures
references/core_workflow.md
Complete output templates, execution rules, extra-long content processing rules, exception handling scripts

常见问答

Frequently Asked Questions

安装相关问题

Installation-related Questions

Q1: 安装时提示缺少依赖库怎么办？
A: 按以下命令安装缺失的依赖：
bash
pip install python-docx==1.1.0 beautifulsoup4==4.12.3 playwright==1.58.0
playwright install chromium
Q2: Playwright 安装失败怎么办？
A: 网页检测需要 Playwright 渲染 JS 页面。如果 Playwright 不可用，脚本会自动回退到静态页面提取方式，不影响其他功能使用。
Q1: What should I do if the installation prompts missing dependency libraries?
A: Install the missing dependencies using the following command:
bash
pip install python-docx==1.1.0 beautifulsoup4==4.12.3 playwright==1.58.0
playwright install chromium
Q2: What should I do if Playwright installation fails?
A: Webpage detection requires Playwright to render JS pages. If Playwright is unavailable, the script will automatically fall back to static page extraction, which does not affect other functions.

使用相关问题

Usage-related Questions

Q3: 一次最多能检测多少内容？
A: 单次检测上限 3000 字符。3000~10000 字符支持分批检测后自动合并，超过 10000 字符建议手动分批。
Q4: 支持哪些文件格式？
A: 支持 TXT、DOC、DOCX、CSV、MD 等文本类型文件，以及常见图片格式（提取图中文字）。暂不支持 PDF，请转为图片或文本文件后上传。
Q5: 数据会外传吗？如何保障隐私？
A: 文案将通过加密 HTTPS 连接发送至检测服务进行匹配，不在本地存储。
Q3: What's the maximum content that can be detected at one time?
A: The single detection limit is 3000 characters. For 3000~10000 characters, batch detection and automatic result merging are supported. For content exceeding 10000 characters, manual batch processing is recommended.
Q4: What file formats are supported?
A: Supports text-type files such as TXT, DOC, DOCX, CSV, MD, as well as common image formats (extract text from images). PDF is not supported temporarily, please convert it to image or text file before uploading.
Q5: Will data be transmitted externally? How to ensure privacy?
A: The copy will be sent to the detection service via encrypted HTTPS connection for matching, and will not be stored locally.

故障排除

Troubleshooting

Q6: 检测服务返回异常怎么办？
A: 脚本内置自动重试机制（最多 3 次），偶尔网络波动会自动重试。如果多次重试仍失败，请稍后重试或联系技术支持。
Q7: 英文单词被误判为违禁词？
A: 已内置英文误匹配过滤。脚本会自动识别并过滤英文单词内部子串误标（如 "Glasswing" 中的 "ass"），无需额外处理。
Q8: 文件或网页提取失败？
A: 请检查文件格式是否正确、网址是否可正常访问。PDF 文件不支持，请先转换为图片或文本格式。
Q6: What should I do if the detection service returns an exception?
A: The script has a built-in automatic retry mechanism (up to 3 times). Occasional network fluctuations will trigger automatic retries. If retries fail multiple times, please try again later or contact technical support.
Q7: Are English words misjudged as prohibited words?
A: English false match filtering is built-in. The script will automatically identify and filter false marking of substrings inside English words (e.g., "ass" in "Glasswing"), no additional processing is required.
Q8: What should I do if file or webpage extraction fails?
A: Please check if the file format is correct and if the website is accessible. PDF files are not supported, please convert them to image or text format first.

获取帮助

Get Help

  • 完整执行规程、输出模板、执行铁律详见 
    references/core_workflow.md
  • Complete execution procedures, output templates, and execution rules are detailed in 
    references/core_workflow.md

触发条件

Trigger Conditions

满足以下任一即激活本 Skill：
  • 用户提到「抖音违禁词」「抖音敏感词」「抖音审核」「限流词」「脚本合规」「口播文案合规」
  • 用户要求检测抖音短视频脚本、直播话术、字幕文案、商品详情等内容安全性
  • 用户上传文件/图片/链接并提到抖音相关的检测需求
This Skill will be activated if any of the following conditions are met:
  • User mentions "Douyin prohibited words", "Douyin sensitive words", "Douyin review", "restriction words", "script compliance", "voiceover copy compliance"
  • User requests to detect the safety of Douyin short video scripts, live stream scripts, subtitle copy, product details, etc.
  • User uploads files/images/links and mentions Douyin-related detection needs

执行铁律

Execution Rules

调用脚本、格式化输出前，必须先读取并严格遵循 
references/core_workflow.md
 全文。核心原则：
  1. 禁止编造结果 —— 仅以脚本返回 JSON 为准，不得臆造违禁词或替换建议
  2. 禁止输出原始 JSON —— 必须解析后填入输出模板
  3. 禁止多余寒暄 —— 仅输出三个板块（🔍检测结果、💡修改建议、✏️优化文案），无开场白/结束语
  4. 分批/字数超限必须暂停 —— 发出提示后立即停止，等待用户明确回复 1/2/3 后再继续
  5. 图片仅限文字 —— 禁止获取图片的视觉风格、布局、色彩、人物等任何非文字信息
  6. 必须交付文件 —— 三板块输出完成后，必须将纯文本优化文案写入 
    ./抖音_优化文案_{随机6位数字}.txt
    并以卡片形式发送给用户，禁止省略 A: 已内置英文误匹配过滤。脚本会自动识别并过滤英文单词内部子串误标（如 "Glasswing" 中的 "ass"），无需额外处理。
Q8: 文件或网页提取失败？
A: 请检查文件格式是否正确、网址是否可正常访问。PDF 文件不支持，请先转换为图片或文本格式。
Before calling the script and formatting the output, must read and strictly follow the full text of 
references/core_workflow.md
. Core principles:
  1. Forge no results —— Only rely on the JSON returned by the script, do not fabricate prohibited words or replacement suggestions
  2. Do not output raw JSON —— Must parse it and fill it into the output template
  3. No unnecessary greetings —— Only output the three sections (🔍Detection Results, 💡Revision Suggestions, ✏️Optimized Copy), no opening/closing remarks
  4. Must pause for batch/word count exceeding limits —— Stop immediately after sending the prompt, wait for the user's clear reply of 1/2/3 before continuing
  5. Images are for text only —— Forbid obtaining any non-text information such as visual style, layout, color, characters of the image
  6. Must deliver files —— After outputting the three sections, must write the plain text optimized copy to 
    ./Douyin_Optimized_Copy_{random 6-digit number}.txt
    and send it to the user as a card, forbidding omission

获取帮助

—
  • 完整执行规程、输出模板、执行铁律详见 
    references/core_workflow.md
—

触发条件

—
满足以下任一即激活本 Skill：
  • 用户提到「抖音违禁词」「抖音敏感词」「抖音审核」「限流词」「脚本合规」「口播文案合规」
  • 用户要求检测抖音短视频脚本、直播话术、字幕文案、商品详情等内容安全性
  • 用户上传文件/图片/链接并提到抖音相关的检测需求
—

执行铁律

—
调用脚本、格式化输出前，必须先读取并严格遵循 
references/core_workflow.md
 全文。核心原则：
  1. 禁止编造结果 —— 仅以脚本返回 JSON 为准，不得臆造违禁词或替换建议
  2. 禁止输出原始 JSON —— 必须解析后填入输出模板
  3. 禁止多余寒暄 —— 仅输出三个板块（🔍检测结果、💡修改建议、✏️优化文案），无开场白/结束语
  4. 分批/字数超限必须暂停 —— 发出提示后立即停止，等待用户明确回复 1/2/3 后再继续
  5. 图片仅限文字 —— 禁止获取图片的视觉风格、布局、色彩、人物等任何非文字信息
  6. 必须交付文件 —— 三板块输出完成后，必须将纯文本优化文案写入 
    ./抖音_优化文案_{随机6位数字}.txt
    并以卡片形式发送给用户，禁止省略
—