WhatsHot
You can discover WhatsHot sources, fetch current boards, and analyze stored historical trends.
List all discoverable WhatsHot sources without network access.
Inspect a source’s boards and accepted parameters.
Fetch current hotlist, newsflash, or gold boards from upstream without saving history.
Query persisted historical items with filters and cursor pagination.
Search stored titles, descriptions, and newsflash content with attributable evidence.
Get rank and hot-value trend series for a specific historical item.
Check local history storage counts and freshness.
Fetches trending hot lists from Bilibili via the WhatsHot API, CLI, and MCP tools.
Serves as the local persistent history database for the WhatsHot service, storing captured data for historical queries, search, and trend analysis.
The underlying Python web framework powering the WhatsHot API service.
Provides optional shared response caching across multiple API instances.
Aggregates RSS and RSSHub feeds and can output data as RSS 2.0 feeds from the same endpoints.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@WhatsHot what are the top trending topics on Weibo right now?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
🔥 WhatsHot · 什么火了
一个工具,看遍全网热榜。
微博、知乎、抖音、B 站、GitHub、财联社、周大福…… 把 400+ 个网站的热搜、实时快讯和金价整理成同一种格式,随取随用。
在线体验 · 三分钟上手 · 支持哪些网站 · 常见问题 · 开发者文档
基于本项目构建的网页版 whatshot.top
为什么做这个
热点散落在几十个 App 和网站里:微博看热搜,知乎看讨论,B 站看视频,财联社盯快讯, 金店官网查金价……每个都要单独打开,格式也各不相同。
WhatsHot 把它们收拢到一处。每个来源都会变成结构相同的列表,包含标题、链接、热度和时间。 你可以在终端里直接看,用 RSS 阅读器订阅,交给 AI 助手分析,也可以接进自己的产品。
只想看看热榜,不想安装?直接打开 whatshot.top,网页版基于本项目构建。
Related MCP server: TrendRadar
你可以用它来
场景 | 怎么做 | |
📰 | 每天花一分钟扫一遍全网热点 | 终端里输入 |
📡 | 用 RSS 阅读器订阅任意榜单 | 榜单地址后加 |
🤖 | 让 AI 帮你查热点、追事件、看趋势 | 配合 whatshot-mcp 接入 AI 助手 |
💰 | 对比各家金店今天的金价 | 周大福、周生生、老凤祥、六福等金价榜 |
🧩 | 做热点看板、群聊机器人、每日简报 | 调用 JSON 接口 |
🕰️ | 回看某个话题何时上榜、热了多久 | 开启定时采集,保存历史 |
三分钟上手
1. 安装 uv(Python 包管理工具,会自动准备 Python 3.12+)
# macOS / Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
# Windows(PowerShell)
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"2. 下载项目
git clone https://github.com/alisen39/whatshot.git
cd whatshot
uv sync3. 看一眼微博热搜
uv run whatshot weibo --limit 5 -f plain输出类似(示例):
1. 某地出现罕见天象
2. 新款手机今日开售
3. 周末全国大范围降温
4. 某综艺官宣新嘉宾
5. 高校食堂推出新菜品换成别的网站,只需要换一个名字:
uv run whatshot zhihu -f plain # 知乎热榜
uv run whatshot bilibili -f plain # B 站全站排行
uv run whatshot cls -f plain # 财联社电报
uv run whatshot --help # 查看全部网站想要别的格式,把 -f plain 换成 json、markdown 或 csv。
4. 启动接口服务(可选)
uv run python -m whats_hot_api然后在浏览器里打开:
地址 | 内容 |
微博热搜(JSON) | |
同一个榜单的 RSS 订阅地址 | |
微博有哪些榜单可选 | |
全部网站列表 |
把 RSS 地址填进 Inoreader、Feedly、NetNewsWire、Folo 等阅读器,就能像订阅公众号一样订阅热榜。
在服务器上部署时,把 127.0.0.1 换成服务器地址即可。
支持哪些网站
目前收录 400+ 个网站、2000+ 个榜单,分为热榜、快讯、金价三类。下面只列一部分:
类别 | 代表来源 |
综合热搜 | 微博、知乎、百度、抖音、快手、今日头条、百度贴吧、豆瓣、虎扑、Google Trends |
视频与音乐 | 哔哩哔哩、AcFun、腾讯视频、爱奇艺、猫眼电影、网易云音乐、QQ 音乐、YouTube、TikTok、Spotify |
新闻 | 澎湃新闻、新华网、人民网、央视网、腾讯新闻、网易新闻、联合早报、BBC、Reuters、纽约时报 |
科技与开发者 | 36氪、虎嗅、少数派、IT之家、V2EX、稀土掘金、GitHub、Hacker News、Product Hunt、Stack Overflow |
AI 与科研 | OpenAI、Anthropic、Google DeepMind、Hugging Face、arXiv、Nature、Science |
财经快讯 | 财联社、华尔街见闻、金十数据、东方财富、同花顺、第一财经、雪球 |
游戏与 ACG | Steam、NGA、米游社、原神、崩坏:星穹铁道、英雄联盟、游研社 |
生活与阅读 | 什么值得买、微信读书、豆瓣读书、喜马拉雅、懂球帝、中央气象台、历史上的今天 |
金价 | 周大福、周生生、老凤祥、六福珠宝、中国黄金、菜百首饰、周六福 |
完整清单:运行 uv run whatshot --help,或启动服务后访问 /all。
更多用法
接入 AI 助手
whatshot-mcp 是配套的 MCP 服务。接入 Claude、 Cursor 等支持 MCP 的 AI 工具后,可以直接问它“今天各平台都在讨论什么”“这件事最早在哪儿上的榜”。 它既可以连接你自己运行的 WhatsHot,也可以连接 WhatsHot 托管服务。
定时采集与历史数据
默认只返回“此刻”的榜单。如果想知道某个话题什么时候上榜、热了多久,可以启动后台采集服务, 按你设定的频率保存到本地数据库,之后再按关键词搜索、看趋势。配置方法见 定时采集与历史数据。
常见问题
需要注册账号或申请 API Key 吗? 不需要。数据来自各网站公开的页面和接口,直接运行即可。
一定要装 Redis 或数据库吗? 不需要。没有 Redis 时自动只用内存缓存;只有开启历史采集时才会用到本地 DuckDB 文件,无需单独安装。
数据多久更新一次?
每次请求时实时抓取,并缓存一段时间避免频繁访问:热榜默认 30 分钟,快讯默认 5 分钟。
需要最新数据时,在地址后加 ?cache=false,或在命令行加 --cache refresh。
海外网站抓不到?
GitHub、纽约时报等部分海外来源在国内网络需要代理,可以在 .env 里用 ROUTE_PROXY 为指定域名配置代理,
见 HTTP API 与配置。
某个网站突然不能用了? 网站改版很常见。欢迎提交 Issue 告诉我们,或者直接提 PR 修复。
数据可以商用吗? 代码使用 MIT 许可,可以自由使用。抓取到的内容版权归各来源网站所有,请遵守对方的服务条款, 并合理控制请求频率。
开发者文档
HTTP API 与配置:接口格式、查询参数、金价字段、缓存与代理配置、命令行
定时采集与历史数据:daemon、DuckDB 历史库、Backend Contract、MCP 对接
开发与扩展:新增网站、快讯回溯声明、作为库嵌入、测试
想新增一个网站?通常只需要在 whats_hot_api/routes/ 下加一个文件,见
开发与扩展。
License
MIT。第三方代码授权说明见 THIRD_PARTY_NOTICES.md。
Available Tools
7 toolsfetch_currentBRead-onlyIdempotent
Fetch a current board. This may access the upstream website but never persists a history capture.
Inputs: site (required), limit, params, path_type. Returns: kind, site, type, items, +5 more
get_source_schemaBRead-onlyIdempotent
Describe one source, its boards, and accepted parameters.
Inputs: site (required). Returns: source
get_storage_statsARead-onlyIdempotent
Return local history storage counts and freshness.
Inputs: no named parameters. Returns: enabled, captures, goldRows, hotlistRows, +2 more
get_trend_seriesCRead-onlyIdempotent
Get rank and hot-value trends for one historical item.
Inputs: site (required), item_id (required), board_key (required), since, +2 more. Returns: site, bucket, itemId, series, +1 more
list_sourcesARead-onlyIdempotent
List all discoverable WhatsHot sources without network access.
Inputs: no named parameters. Returns: sources
query_historyCRead-onlyIdempotent
Query persisted historical items with cursor pagination.
Inputs: kind, site, limit, since, +3 more. Returns: items, truncated, nextCursor
search_historyBRead-onlyIdempotent
Search persisted titles, descriptions, and newsflash content. Returns attributable evidence, not free-form summaries.
Inputs: keyword (required), site, limit, since, +3 more. Returns: items, truncated, nextCursor
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 7 tools
Each tool targets a distinct concern: source discovery, schema introspection, live fetching, history querying, searching, trend analysis, and storage stats. The main potential confusion is between query_history and search_history, but the descriptions make clear that querying returns persisted items with pagination while searching targets content and returns attributable evidence, which differentiates them sufficiently.
Names mostly follow a verb_noun pattern with clear action prefixes (list_, get_, fetch_, query_, search_, get_). There's a minor deviation with get_source_schema vs fetch_current where 'fetch' and 'get' use different verbs for similar acquisition actions, but this is arguably intentional given that fetch_current accesses the network while get_source_schema is local.
Seven tools is within the ideal 3-15 range and each tool appears to earn its place by covering a distinct dimension of the domain: discovery, introspection, live fetch, history retrieval, search, analytics, and storage management. No redundant or filler tools.
The surface covers the full lifecycle: source discovery, schema understanding, live fetching, persistence querying, search, trend analytics, and storage management. Minor gaps exist—such as the absence of explicit history write/capture control (though fetch_current implies an upstream capture mechanism) and possibly pagination/sampling controls for trend series—but no obvious dead ends that would cause agent failures.
Maintenance
Related MCP Connectors
MCP server aggregating hot-search boards from 8 Chinese platforms (Weibo, Zhihu, Bilibili, Douyin).
MCP server for real-time Chinese internet hot-search and trending boards across major platforms.
One MCP server for all web data. Scrape, search, crawl, and AI research in one install.
Unified 15-tool SEO, SERP, clean Markdown & GEO (LLM citation Share-of-Voice) MCP server.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceA modern multi-platform hot news aggregation server based on FastMCP that supports real-time hot topics data collection from 13+ major platforms including Zhihu, Weibo, Baidu, and Bilibili.3MIT
- AlicenseAqualityDmaintenanceAn aggregator for real-time hot topics and news from major social and financial platforms like Zhihu, Bilibili, and Wall Street News. It features an MCP server that allows AI models to fetch and analyze trending information for automated insights.17GPL 3.0
- AlicenseNot gradedqualityDmaintenanceA Python-based MCP server that aggregates real-time trending topics and news from over 30 sources including social media, technology, and finance platforms. It enables AI applications to retrieve the latest hot searches and news from major services like Weibo, Baidu, and IT Home through a unified interface.33MIT
- FlicenseAqualityDmaintenanceMCP server for fetching daily hot lists from 30+ sources with built-in caching and batch requests.56 npm1-