知乎热榜 & 知乎 API | Zhihu Scraper (知乎爬虫) & MCP Server

Zhihu Scraper is a ready-made Zhihu API for Zhihu (知乎), China's largest Q&A platform. Get the live Zhihu hot list (知乎热榜 / 知乎热搜 / zhihu rebang) and run search, answers, articles, comments, creator profiles, topics and Zhuanlan columns (知乎专栏), all with no login, cookie or API key. Export to JSON, CSV or Excel, get Slack, Telegram or Discord alerts when a topic trends, or connect AI agents through the built-in Zhihu MCP server.

What can this Zhihu scraper do?

• 🔥 Zhihu hot list (知乎热榜): all 9 boards with rank, heat score (热度), excerpt, answer and follower counts
• 📈 Hot list tracking: each run compares with the previous one, giving rank_change, heat_change and new / up / down / dropped status
• 🔔 Trend alerts: get pinged in Slack, Telegram or Discord when your keyword hits the hot list
• 🔎 Zhihu search: answers, articles and videos with type, sort and date filters, plus search suggestions, topic search and column search
• 💬 Answers and comments: every answer to a question (full text, optional Markdown), comments and replies
• 👤 Creators: profiles (followers, upvotes, verification, employer, IP location), their answers and articles
• 📚 Zhuanlan (知乎专栏): column search and every article in a column
• 🤖 Zhihu MCP server: Claude, Cursor and other MCP clients can query Zhihu live
• 🚫 No login needed: nothing to set up; optionally use your own Zhihu account or official API key for unlimited runs

中文说明

知乎爬虫 / 知乎 API：知乎热榜（全站、科学、数码、体育、时尚、影视、校园、汽车、深度 9 个榜单）、知乎搜索、问题回答、文章详情、知乎评论及回复、用户资料、话题和知乎专栏文章，无需登录、无需 Cookie 即可采集。也可以填写你自己的知乎开放平台 Access Secret 或 Cookie，不受每日额度限制。

• 热榜监控：记录排名与热度变化（上升、下降、新上榜、掉榜），支持关键词提醒，推送到 Slack、Telegram 或 Discord。
• 数据导出：JSON、CSV、Excel，或通过 Apify API 和 Webhook 接入你的系统。
• AI 接入：内置 MCP Server，可在 Claude、Cursor 等 AI 工具中直接查询知乎。
• 知乎开放平台：Access Secret 每天 1000 次免费调用，用于搜索和问题回答；Cookie 可解锁全部功能。
• 按结果计费：每条 $0.002，失败不收费。

Use cases

• China social media monitoring: track Chinese public opinion and trending topics hour by hour
• Brand and competitor tracking: alerts when your brand trends, plus search and comment exports for sentiment analysis
• Zhihu datasets for AI: full answer text and comments in Chinese for NLP, LLM training and RAG
• KOL and creator research: find experts by topic, compare followers and upvotes
• Zhihu in English: export Chinese answers to CSV or JSON for your own translation or analysis pipeline

How to scrape Zhihu

1. Click Try for free and pick an operation (the Zhihu hot list is the default).
2. Add keywords (for search) or Zhihu URLs (for answers, comments, profiles and columns).
3. Click Start, then download the data as JSON, CSV or Excel, or fetch it through the Apify API.
4. Optional: schedule the run hourly and add a Slack, Telegram or Discord target for a Zhihu trend alert bot.

Operations

operation	Input	Output
hot_list (default)	categories	One row per trending question (知乎热榜)
search	keywords	Answers, articles, videos and questions (知乎搜索)
search_suggestions	keywords	Zhihu's autocomplete suggestions
topic_search	keywords	Topics (话题): name, followers, questions
column_search	keywords	Zhuanlan columns (专栏): title, articles, followers, author
web_search	keywords	Zhihu's web-wide search (needs your official API key)
question_answers	question urls	Every answer, with text
content_detail	answer / article / question urls	One detailed row per URL
comments	answer urls	Comments (知乎评论), plus replies with includeReplies
comment_replies	answer urls + commentIds	Replies to specific comments
user_profile	zhihu.com/people/… urls	Followers, answers, articles, upvotes, verification, employer, IP location
user_articles / user_answers	zhihu.com/people/… urls	A creator's articles or answers, newest first
column_articles	zhuanlan.zhihu.com/<column> urls	Every article in a column

Paging is automatic up to maxItems per keyword or URL. On the hosted backend, content_detail for an answer has no vote or comment counts (the source doesn't provide them); search, question_answers and creator posts do include them, and so does content_detail with your cookie.

Input example

{
  "operation": "hot_list",
  "categories": ["total", "digital"],
  "titleKeywords": ["AI", "华为"],
  "slackWebhookUrl": "https://hooks.slack.com/services/…"
}

Output example (hot list)

{
  "category": "total",
  "rank": 3,
  "title": "如何评价……？",
  "question_url": "https://www.zhihu.com/question/…",
  "heat": 26340000,
  "heat_text": "2634 万热度",
  "answer_count": 812,
  "follower_count": 3021,
  "prev_rank": 7,
  "rank_change": 4,
  "list_status": "up",
  "scraped_at": "2026-10-26T03:00:00+00:00"
}

Zhihu hot list monitoring

Field	Default	Notes
categories	["total"]	total, science, digital, sport, fashion, film, school, car, depth
maxItemsPerCategory	50	1–50
trackChanges	true	Adds prev_rank, rank_change (positive = climbed), heat_change, list_status
includeDropped	false	Also output questions that left the board (list_status: dropped)
titleKeywords	–	Only output questions whose title or excerpt mentions one of these words
onlyNew	false	Skip questions returned in earlier runs
stateKey	default	Separate memory per monitor

Output fields

• Hot list: category, rank, question_id, title, question_url, excerpt, heat, heat_text, answer_count, follower_count, comment_count, label, is_debut, trend, topic_ids, thumbnail_url, created_at, scraped_at, is_new, prev_rank, rank_change, heat_change, list_status, matched_keywords
• Search, answers, articles: operation, record_type, keyword, search_rank, content_type, content_id, title, url, question_id, question_url, excerpt, content, content_markdown, voteup_count, comment_count, answer_count, follower_count, visit_count, play_count, topics, image_url, created_at, updated_at, ip_location, author_*
• Comments: record_type (comment | reply), comment_id, parent_comment_id, content_url, content, like_count, reply_count, reply_to_author, is_author, ip_location, created_at, author_*
• Topics and columns: topic_id, name, url, introduction, follower_count, question_count / column_id, title, url, description, article_count, follower_count, author_*

Every row also has backend, which says how it was fetched.

Zhihu MCP server for AI agents

This actor also runs as a remote Zhihu MCP server (streamable HTTP):

URL:    https://molded_mountainpass--zhihu-hot-list-scraper.apify.actor/mcp
Header: Authorization: Bearer <your Apify API token>

Tools: zhihu_hot_list, zhihu_search, zhihu_search_topics, zhihu_search_columns, zhihu_question_answers, zhihu_content, zhihu_comments, zhihu_user, zhihu_user_posts, zhihu_column_articles. Each returned row is billed like a dataset row.

No login, or your own account

Everything works with no login: the hot list comes straight from Zhihu, and the other operations run on our hosted backend. Hosted capacity is shared and limited per day; when it's used up, runs ask you to try again after 00:00 UTC. Free Apify plans include 5 hosted requests to try it (each request returns up to about 20 results); paid plans aren't trial-limited.

For unlimited runs, add your own account instead (it's used first when present):

• Official Zhihu API key (知乎开放平台): every account gets 1,000 free calls a day. Sign up at developer.zhihu.com, copy the Access Secret from 个人中心 (personal center), and paste it into Zhihu API key. It's used for search (up to 10 results per keyword), question_answers (summaries) and web_search (up to 20). No account-ban risk.
• Zhihu cookie: log in to zhihu.com, open DevTools → Network, click any zhihu.com request and copy the full cookie request header into Zhihu cookie (must contain z_c0). It unlocks every operation on your account, including comments on articles and questions. Keep volumes reasonable; Zhihu may rate-limit heavy use. If a run reports ERR_TICKET_NOT_EXIST, copy a fresh cookie.

Both fields are stored encrypted and only used for your runs. If you add both, the API key is used where it applies and the cookie for the rest. Comments on articles and questions (not just answers) need the cookie.

Notifications

Field	Notes
slackWebhookUrl	Slack incoming webhook
discordWebhookUrl	Discord channel webhook
telegramBotToken + telegramChatId	Bot from @BotFather (add it to the chat first)
notificationLimit	Items per message (default 10)
webhookUrl	All results POSTed as JSON (Make, Zapier, n8n)

Pricing

$0.002 per result (pay per event): $2 per 1,000 rows. Failed lookups produce no rows and cost nothing beyond the small platform start fee.

FAQ

Is there an official Zhihu API? Since 2026 Zhihu has an open platform (知乎开放平台) with a free tier covering search, the hot list and answer summaries. This actor covers much more (comments, creators, columns, full answers) with no account needed, and can also use your official key.

How often does the Zhihu hot list (知乎热榜) update? Every few minutes. Schedule this actor hourly with trackChanges to see which questions are climbing, or with onlyNew to capture each new trending question once.

Can I scrape Zhihu answers and comments without logging in? Yes. Use question_answers, content_detail or comments. Answer text comes as plain text, and as Markdown with includeMarkdown.

Can I read Zhihu in English? Zhihu content is in Chinese. Export it as CSV or JSON and run it through your translation tool or LLM of choice.

Can I use Zhihu data in Claude, Cursor or another AI agent? Yes. Add the MCP server URL above to your MCP client config.

Is scraping Zhihu legal? This actor collects publicly available data. You're responsible for using it in line with Zhihu's terms and applicable laws, such as China's PIPL and the GDPR for personal data.

For developers

pip install -r requirements.txt
python -m src                                  # run locally
pip install pytest && python -m pytest tests   # tests
apify push                                     # deploy