Content Compliance Checker

Overview

This skill enables automated detection of banned words and non-compliant content in self-media platform text. It analyzes titles, copywriting, and other text content against a comprehensive database of banned words from major Chinese self-media platforms (Douyin, Xiaohongshu, WeChat Official Accounts, Bilibili, Kuaishou), then provides specific replacement suggestions for each violation found.

When to Use This Skill

Use this skill when:
- User submits text content (titles, descriptions, articles, marketing copy) for compliance checking
- User asks to verify if content contains banned words or sensitive terms
- User needs replacement suggestions for non-compliant content
- User wants to prepare content for publishing on Chinese self-media platforms
- User mentions "违禁词检测", "敏感词检查", "内容合规检查", "文案审核"

Core Capabilities

1. Comprehensive Banned Word Detection

Detects violations across 9 major categories:
- Absolute terms: "最", "第一", "唯一", "顶级", etc.
- False promises: "100%有效", "无效退款", "零风险", etc.
- Medical claims: "治疗", "根治", "美白", "减肥", etc.
- Superstition: "招财进宝", "旺财旺运", "逢凶化吉", etc.
- Inducement marketing: "秒杀", "抢疯了", "限时最后一天", etc.
- Political sensitive: Government, military, sovereignty-related terms
- Discrimination: Ethnic, racial, gender discriminatory language
- Vulgar content: Profanity and inappropriate language
- Platform-specific: Platform-unique violations (Douyin live streaming, Xiaohongshu external linking, WeChat share inducement)

2. Multi-Platform Coverage

Supports checking against platform-specific rules for:
- Douyin: Live streaming, e-commerce, external linking rules
- Xiaohongshu: External linking, beauty industry, 2026 new terms
- WeChat Official Accounts: Share inducement, follow inducement
- Bilibili: Danmaku, interaction area rules
- Kuaishou: 10 violation dimensions

3. Intelligent Analysis

• Context-aware detection (shows surrounding text for each violation)
• Severity assessment (high, medium, low)
• Compliance scoring (0-100)
• Risk level determination (low, medium, high, critical)

4. Actionable Replacement Suggestions

Provides specific, compliant alternatives for each banned word found, based on the comprehensive replacement database.

Workflow

Step 1: Input Processing

1. Receive user's text content (title, description, article, etc.)
2. Identify which platforms to check against (default: all major platforms)
3. Prepare text for analysis

Step 2: Compliance Checking

1. Run the compliance checker script against the text
2. Detect all banned words across all categories
3. Analyze context for each violation
4. Calculate compliance score and risk level

Step 3: Report Generation

1. Generate detailed violation report
2. Include specific replacement suggestions for each violation
3. Provide overall recommendations
4. Format report according to user preference (text, JSON, or Markdown)

Step 4: User Guidance

1. Present findings to user with clear explanations
2. Highlight highest-risk violations first
3. Offer to help rewrite non-compliant sections
4. Provide general compliance best practices

Usage Examples

Example 1: Basic Compliance Check

User input: "这款产品是全网销量第一的品牌，效果100%有效，无效退款。"

Expected output:
- Detects: "全网销量第一" (absolute term), "100%有效" (false promise), "无效退款" (false promise)
- Provides replacements: "销量领先" (leading sales), "多数用户反馈良好" (good feedback from most users), "支持七天无理由退换" (supports 7-day no-reason return)
- Calculates compliance score and risk level

Example 2: Platform-Specific Check

User input: "加微信获取独家资料，秒杀价最后一天！"

Expected output:
- Detects: "加微信" (Xiaohongshu external linking), "秒杀" (inducement marketing), "最后一天" (time pressure)
- Platform-specific warnings for Xiaohongshu and Douyin
- Suggests: "私信获取" (get via private message), "限时优惠" (limited-time offer), "活动截止日期" (activity end date)

Example 3: Multi-Platform Content

User input: "这款护肤品能根治痘痘，让你美白祛斑，招财进宝！"

Expected output:
- Detects: "根治" (medical claim), "美白" (medical claim), "祛斑" (medical claim), "招财进宝" (superstition)
- Provides beauty industry-specific replacements
- Warns about medical claim regulations

Script Usage

Command Line Interface

# Basic text check
python scripts/compliance_checker.py "要检测的文本内容"

# JSON output format
python scripts/compliance_checker.py "要检测的文本内容" json

# Markdown output format
python scripts/compliance_checker.py "要检测的文本内容" markdown

Python API Usage

from scripts.compliance_checker import ContentComplianceChecker

# Initialize checker
checker = ContentComplianceChecker()

# Check text
report = checker.check_text("要检测的文本内容")

# Generate report
text_report = checker.generate_report(report, "text")
json_report = checker.generate_report(report, "json")
markdown_report = checker.generate_report(report, "markdown")

Reference Database

The skill includes a comprehensive banned words database at references/banned_words_database.md containing:
- 2026 platform audit mechanism updates
- Complete banned words by category (9 major categories)
- Platform-specific violations for 5 major platforms
- Industry-specific violations for 5 industries
- Compliance replacement suggestions
- Self-check lists and tools

Best Practices

1. Check before publishing: Always run compliance check before publishing content
2. Consider context: Some words may be acceptable in certain contexts
3. Use replacements: Don't just delete banned words - replace with compliant alternatives
4. Platform awareness: Different platforms have different rules
5. Regular updates: Banned word lists change frequently - stay updated
6. Avoid variants: Harmonic sounds, pinyin, and symbol variations are also detected

Limitations

1. Context sensitivity: Some violations may be acceptable in specific contexts (e.g., medical articles by certified professionals)
2. Platform evolution: Platform rules change frequently - database may not reflect latest changes
3. Language complexity: Some violations require human judgment for proper assessment
4. Industry specificity: Some industries have additional regulations not covered in general database

Integration

This skill can be integrated with:
- Content management systems
- Social media scheduling tools
- Marketing automation platforms
- Editorial workflow systems
- Quality assurance processes