ai-hot-research

name	ai-hot-research
description	AI-powered multi-platform hot topic monitoring and analysis skill. Fetches real-time trending/hot topics from 5 Chinese platforms: Weibo (微博热搜), Zhihu (知乎热榜), Baidu (百度热搜), Douyin (抖音热点), Bilibili (B站热门). Uses DeepSeek AI (deepseek-chat) for intelligent analysis including: trend clustering, sentiment analysis, single topic deep-dive, and contextual Q&A about current trends. Triggers: (1) "查看热搜/热点/热榜" or "what's trending", (2) "分析热点" or "analyze trends", (3) "查看[平台]热搜" for specific platform, (4) "分析话题 [topic]" for deep analysis, (5) "热点问答" or asking questions about current trends, (6) "热点报告" for comprehensive report. Capabilities: real-time data fetching from 5 platforms (190+ topics), AI clustering & trend analysis, sentiment detection, single topic deep-dive, conversational Q&A with trend context, and formatted report generation.
author	AI Hot Research Team
version	1.0.0
date	"2026-03-04T00:00:00.000Z"

AI Hot Research - 热点监控 Agent Skill

Overview

This skill provides real-time multi-platform hot topic monitoring with AI-powered analysis. It fetches trending data from 5 major Chinese platforms and uses DeepSeek AI to generate intelligent insights, trend clusters, sentiment analysis, and comprehensive reports.

When to Use This Skill

Use this skill when the user:

• Asks about what's trending or hot right now (热搜/热点/热榜)
• Wants to know hot topics on specific platforms (微博/知乎/百度/抖音/B站)
• Requests trend analysis or hotspot clustering (分析热点/趋势分析)
• Wants deep analysis of a specific topic (分析话题/深度分析)
• Asks questions about current trends (热点问答/今天有什么大事)
• Needs a comprehensive hot topic report (热点报告/舆情报告)
• Mentions keywords: 热搜, 热榜, 热点, trending, hot topics, 舆情

Architecture

┌─────────────────────────────────────────────┐
│              Agent Skill Layer              │
│  SKILL.md + scripts/                        │
├─────────────────────────────────────────────┤
│           scripts/fetch_topics.py           │  ← Data fetching
│           scripts/ai_analyze.py             │  ← AI analysis
│           scripts/report.py                 │  ← Report generation
├─────────────────────────────────────────────┤
│  Data Sources (httpx async):                │
│  ┌─────┐ ┌─────┐ ┌─────┐ ┌──────┐ ┌─────┐│
│  │Weibo│ │Zhihu│ │Baidu│ │Douyin│ │Bili │ │
│  └─────┘ └─────┘ └─────┘ └──────┘ └─────┘│
├─────────────────────────────────────────────┤
│  AI Engine: DeepSeek (deepseek-chat)        │
│  via OpenAI-compatible API                  │
└─────────────────────────────────────────────┘

Supported Platforms

Platform	Chinese Name	API Endpoint	Typical Count
Weibo	微博热搜	hot_band / side/hotSearch	50
Zhihu	知乎热榜	api.zhihu.com/topstory/hot-list	30
Baidu	百度热搜	top.baidu.com/api/board	50
Douyin	抖音热点	hot/search/list	30
Bilibili	B站热门	search/square / popular	30

Commands & Usage

1. Fetch Hot Topics (获取热搜)

Fetches real-time trending topics from all or specific platforms.

Trigger phrases: "查看热搜", "今天热搜", "what's trending", "show me hot topics"

Usage:

# Fetch all platforms
python scripts/fetch_topics.py

# Fetch specific platform
python scripts/fetch_topics.py --platform weibo
python scripts/fetch_topics.py --platform zhihu
python scripts/fetch_topics.py --platform baidu
python scripts/fetch_topics.py --platform douyin
python scripts/fetch_topics.py --platform bilibili

# Limit results per platform
python scripts/fetch_topics.py --limit 10

# Output as JSON
python scripts/fetch_topics.py --json

Expected output: List of topics per platform with rank, title, hot value, and URL.

2. AI Trend Analysis (AI 热点分析)

Analyzes current hot topics using DeepSeek AI to identify clusters, trends, and sentiment.

Trigger phrases: "分析热点", "analyze trends", "热点趋势分析"

Usage:

# Full analysis (fetches + analyzes)
python scripts/ai_analyze.py

# Analyze with specific focus
python scripts/ai_analyze.py --focus "科技"
python scripts/ai_analyze.py --focus "finance"

Returns: JSON with clusters (name, keywords, trend, sentiment, heat_score), overview, recommendation.

3. Single Topic Deep Analysis (单话题深度分析)

Deep-dives into a specific topic with background, timeline, public opinion, and prediction.

Trigger phrases: "分析话题 [topic]", "深度分析 [topic]", "analyze topic [topic]"

Usage:

python scripts/ai_analyze.py --topic "2026全国两会"

Returns: JSON with background, timeline, key figures, public opinion (positive/negative/neutral), trend prediction, related topics, impact score, category.

4. Trend Q&A (热点问答)

Ask questions about current trends with AI-powered contextual answers.

Trigger phrases: "关于热点的问题", "hot topic Q&A"

Usage:

python scripts/ai_analyze.py --chat "今天最值得关注的热点是什么？"
python scripts/ai_analyze.py --chat "科技领域有什么新动态？"

5. Comprehensive Report (综合报告)

Generates a formatted markdown report of current hotspots with AI analysis.

Trigger phrases: "热点报告", "生成报告", "generate report"

Usage:

python scripts/report.py
python scripts/report.py --output report.md

Configuration

Required: DeepSeek API Key

Set the DEEPSEEK_API_KEY environment variable or create a .env file:

export DEEPSEEK_API_KEY=sk-your-api-key-here

Or in .env:

DEEPSEEK_API_KEY=sk-your-api-key-here

Optional: Custom Settings

# Change AI model (default: deepseek-chat)
export HOT_RESEARCH_MODEL=deepseek-chat

# Change cache TTL in seconds (default: 300)
export HOT_RESEARCH_CACHE_TTL=300

Dependencies

Required Python packages (pip install):

• httpx - Async HTTP client for fetching platform data
• openai - OpenAI-compatible client for DeepSeek API
• python-dotenv - Environment variable management

pip install httpx openai python-dotenv

Data Source Details

Weibo (微博)

• Primary: https://weibo.com/ajax/statuses/hot_band (免登录, 50条)
• Fallback 1: https://weibo.com/ajax/side/hotSearch (侧边栏)
• Fallback 2: https://m.weibo.cn/api/container/getIndex (移动端)
• Fields: word, num (热度值), category

Zhihu (知乎)

• Primary: https://api.zhihu.com/topstory/hot-list (免登录, 30条)
• Fallback: https://www.zhihu.com/api/v4/search/top_search (热搜词)
• Fields: target.title, detail_text (e.g. "568 万热度")

Baidu (百度)

• Primary: https://top.baidu.com/api/board?tab=realtime (PC端, 50条)
• Fallback: https://top.baidu.com/api/board?platform=wise&tab=realtime
• Fields: word, hotScore, tag

Douyin (抖音)

• Endpoint: https://www.douyin.com/aweme/v1/web/hot/search/list/
• Fields: word, hot_value, word_type

Bilibili (B站)

• Primary: https://api.bilibili.com/x/web-interface/wbi/search/square (热搜词)
• Fallback: https://api.bilibili.com/x/web-interface/popular (热门视频)
• Fields: keyword/title, heat_score/view count

Error Handling

All data sources use fallback strategies:

1. Try primary endpoint
2. On failure, try fallback endpoints
3. Return empty list if all fail (graceful degradation)
4. Print diagnostic messages: [SourceName] method_name 失败: error

Example Outputs

Fetch Output

📊 AI Hot Research - 热点监控
━━━━━━━━━━━━━━━━━━━━━━━━━━━━

🔴 微博热搜 (50 条)
  #1  两会 ..................... 1,066,305
  #2  中国石油发布公告 ......... 775,666
  #3  2026全国两会看点 ......... 588,216

🔵 知乎热榜 (30 条)
  #1  好房子好小区建设 ......... 570 万热度
  #2  腾讯中学生实习 ........... 547 万热度

🟢 百度热搜 (50 条)
  #1  三桶油齐发公告 ........... 7,809,317
  ...

AI Analysis Output

{
  "clusters": [
    {
      "name": "2026年全国两会",
      "keywords": ["两会", "政府工作报告", "代表委员"],
      "trend": "rising",
      "sentiment": "neutral",
      "heat_score": 95,
      "summary": "全国两会成为最热话题...",
      "reason": "年度最重要政治事件"
    }
  ],
  "overview": "当前热点呈现硬新闻与软话题交织...",
  "recommendation": "最值得关注的话题是..."
}

Notes

• All API endpoints are unofficial/reverse-engineered and may change
• Zhihu official API requires authentication; we use api.zhihu.com subdomain which works without auth
• Rate limiting: built-in 15s timeout per request, no aggressive polling
• Cache: 5-minute TTL to avoid excessive API calls
• Data is fetched concurrently for performance (asyncio.gather)

Web Interface

A full web UI is also available at the project path:

/Users/aviator/Documents/code/codefather/AI_Hot_Research/

Start with:

conda activate fastapi
cd /Users/aviator/Documents/code/codefather/AI_Hot_Research
python -m uvicorn backend.main:app --host 0.0.0.0 --port 8000 --reload

Then visit http://localhost:8000 for the Cyberpunk-themed dashboard.

References

• DeepSeek API Documentation
• Weibo Ajax API
• Zhihu API
• Baidu Top
• Bilibili API