Blogs, Knowledge Bases (for tech), Teams Co-edit

AI 知识库工具对比表 (UPDATED 2025-02) #

警告：对于科研文献的理解，暂时只发现 Chat、RAG、summarizer 一类简单的插件，尚未有能够取代人类进行论文研读的插件，可以参考 DeepResearch。对于（论文预讲/详解）类的视频的理解反而有 BibiGPT, VideoSeek 等软件可用。

说明：

1. 计分规则：✅原生支持=1 分｜⚠️插件/第三方支持=0.9 分（括号内标注对应工具名）｜❌不支持=0 分
2. 权重规则：
   • 输入类列（主流办公文档、本地音视频、YouTube 链接、B 站链接、网页剪藏）：单权重=0.5
   • 其余所有功能列：单权重统一为 1，无特殊权重
   • Cost/License列 100% 匹配 alternativeto.net 官方标注，不计入权重与总分
3. 排序规则：按「综合总分」从高到低排序，价格列同步标注对应版本的存储容量限制，数据均来自官方渠道与 alternativeto.net

ref: doubao.

推荐 Obsidian 插件：

• aDHL (依赖 Dynamic Highlights),
• Advanced Tables,
• Collapse All (快捷键建议 C A <),
• Image auto upload (by By renmu, 依赖系统 PicGo 监听),
• Linter,
• Open in Terminal,
• Opener: New Tab by Default,
• Omnisearch (C S F)
• Replace All,
• Sortable Tables,
• Vantage - Advanced search builder (使用 Obsidian 自带搜索栏的高级语法).
• 其它：Global Search and Replace 慎用，容易误操作 (搜索框回车、单击结果都会替换) 且无法撤销。

ps:

• Zotero, EndNote Web,EndNote Web are designed for scholar/academic purpose.
• Logseq 各种操作都需要确认，与“默认执行，可以撤销”的范式不符。

Obsidian for academic work #

WARN: try different ways step by step before fixing your mind (e.g.: max linking for read vs. write and export each chapter as Typst then Latex; vs. do everything in Obsidian … )

• https://www.emilevankrieken.com/blog/2025/academic-obsidian/ , bak
• https://www.youtube.com/watch?v=hSTy_BInQs8
• https://effortlessacademic.com/best-obsidian-plugins-for-academics/
• https://www.reddit.com/r/ObsidianMD/comments/1c81ih1/comment/l0c2vro/?utm_source=share&utm_medium=web3x&utm_name=web3xcss&utm_term=1&utm_content=share_button
• https://github.com/MariosGkMeng/Straightforward-Obsidian2Latex

Video Clipper #

• BibiGPT/VideoSeek 浏览器插件 + 桌面 APP 一键插入 Obsidian
• obsidian小技巧：b站视频时间戳笔记生成（media extended）+信息自动获取 , (文字版)

Web Clipper Archiving & Sharing 作为参考资料的网页剪藏和分享 #

在 1~2 年的实践中，大概有十分之一甚至更多参考资料的网页消失，有些是整个网站消失，因此作为参考资料时，网页备份很重要。

• Evernote Web Clipper，曾经最适合剪藏的，现在共享功能有问题（需登录才能看，故意的？）
• Joplin，剪藏的网页格式有问题，剪藏到本地 Joplin 速度非常快，本地 Joplin 可以网页分享（200 RMB/Y）

UPDATED 2023 #

See also: Project & Code Management Platform Comparison

Personal Blogs #

• Public host 【not recommended due to the lack of desktop clients】: CSDN / CNBlogs (博客园) / zhuanlan.zhihu (知乎专栏) / wordpress.com / blogger / …
• Self-host/Private host framework (LAMP/XAMPP/MEAN): wordpress.org / Drupal / Joomla / Hugo / Farbox2 …
• Public/Private host: Hexo/JekyII/Hugo + github.io / gitee.io / …
  • Near real time solution: Farbox (open source version of Bitcron), Docsify (filename for URL, not good for SEO), hugo server.

Note:

LAMP = Linux, Apache, MySQL and PHP/Python/Perl. XAMPP = cross platform, Apache, MySQL/MariaDB, PHP/Python/Perl. MEAN = MongoDB, Express (for Node.js), AngularJS, Node.js.

Knowledge Base, Knowledge Management, Online Office Recommendations #

先说结论：

• 息流真正做到了 Block 块级的双链（且支持 pdf 双链），但是可能正是因为这个原因，和现有的 Word 等兼容度极差。但是对于技术流、会编程（尤其是 jupyter）的人，非常友好。
• 飞书和钉钉文档各有优势。飞书的思维方式其实更高一筹，但是钉钉（阿里）的整合能力实在太强，飞书面对阿里有点看星星多却看路少的感觉，对新的小团体（<50）用户吸引力略逊色。
• 腾讯文档适合日常办公，适合兼容非技术的行政人员，不适合知识共享。
• 提示：以知识共享为目的的软件，一般都不支持导入 PPT（这个还得靠腾讯）。

各自优劣势（后边进行详细价格和功能比较）：

✔ 本地 markdown.md + vscode （可以配合自建或托管 GitHub 等在线博客） ✔ QQ 腾讯文档: 在线实时协作编辑, 兼容 office，支持批注导出.腾讯的普通文档不占空间,金山所有文档占用空间且默认自动上传打开的所有文件. ✔ 飞书 Feishu: 整体非常优秀. 贵,适合团队高级阶段 (管文化阶段) 内容产出. 在线实时协作编辑, 不支持批注导出。 ✔ 钉钉文档 DingTalk: 整体优秀. 价格合适,适合团队前期 (管人、管物阶段). 导入 docx 时不支持 URL，各级标题格式支持不好 ✔ 我来 Wolai（被阿里收购）：适合笔记管理：块级编辑器 + 支持双链, 不支持批注导出。 ✔ 辅助插件 Web Clipper by DiamondYuan: 保存网页图片等，支持直连语雀等，支持直连 Github(.md). ✔ OneNote by MicroSoft: 强大, 适合个人知识管理, 但是网页格式和对图片中文搜索不友好。 乄 FLowUS 息流: 真 block 级别各类双链，但是几乎不兼容 Office. 支持 PDF 标注嵌入文档. 支持导出为 PDF、 Markdown & CSV、 Word. 乄 石墨文档: 在线实时协作编辑, 不支持批注导出。 ? Siyuan 思源笔记. 开源,适合个人,多端同步功能付费 × Yuque 阿里语雀：适合团队知识管理，对更大、更长的文档更友好。 但是本身 edit-publish 模式较慢，不适合短文档或实时编辑，没有原生 app？ ? 有道云笔记: 过于简单, 据说有不少同步错误导致丢失、搜索很差. ? flomo 浮墨笔记. × Notion: 太复杂，不如直接用语雀（加上网络问题、全英文） × Confluence (by Atlassian) （类似阿里语雀，虽然有中文，除非用 Atlassian 比较习惯，否则没必要）. × Evernote/印象笔记: 当 archive 可以，比较稳定，尤其是网页格式，是除了 SingleFile 之外，格式保存最正确的；但是除了网页保存功能外，其他功能非常不实用, 另外用户体验很差。

ref: Notion, 语雀, 我来(wolai.com)的比较 Notion、语雀、印象笔记、飞书、石墨、有道云笔记，六者之间做出了什么选择，为什么这么选？ 语雀 vs 我来

Feishu 飞书 #

Con:

• no pdf annotations.
• proj. mgmt. module requires individual payments.
• max size of exportable file: 100MB
• canNOT export markdown (only word/pdf) directly, seems API can.
• tech limits ref

free + verified version

price comparison

DingTalk 钉钉 #

Tip: max extra paid space for Std. version: 300GB, 1000RMB/100GB/year.

Tech. advantage: vertical 2-column split view.

Tech. limits:

• import file size: 20MB.
• embedded file size: 8GB.
• chat file size: 4GB.
• chat file hist.:6 mon. (can be moved/synced to 钉盘 for persistancy)
• ref

FlowUS 息流 #

Tip: by using personal pro version, one can invite unlimited collaborators.

price ref

Yuque 阿里语雀 #

free personal version limits #

File storage: 10 GB. Single documents/repository collaborators: 10. Single image upload: 10MB. Single basic file upload: 100MB. Video/compressed files/etc. upload: not available. ref

free (life-time) team version limits #

File storage: 10 GB. Members: 50. Single image upload: 10MB. Single basic file upload: 100MB. Video/compressed files/etc. upload: not available. ref

limits compraed to paied team version #

Operation/access log: not available. External contacts: not available. Sharing management: not available. Data analysis: not available. API calling: not available. Full Text Search: available. AI translation: available. ref

Confluence (by Atlassian) #

• Remote-Friendly Team Workspace.
• Confluence is purpose-built for teams that need a secure and reliable way to collaborate on mission-critical projects.
• Chinese version available (in Atlassian account settings).

free version limits #

2 GB file storage. 10 users. 1 site == 1 second-level domain name. With Mobile App. Integrated with Jira. 1 space = 1 team. 1 personal space = 1 person. Pages: ineractive proj plans, blogs, logs… ref

Hugo #

install #

Due to getting a DBus error with snap, apt is suggested. Search e.g. “amd64.deb” on https://github.com/gohugoio/hugo/releases , right-click, copy link.

Tip 1: many things are changed since v0.124 (2024-03). Tip 2: hugo is enough, hugo-extended is from new templates with new SASS, PostCSS etc. “deploy” is for AWS cloud etc.

wget https://github.com/gohugoio/hugo/releases/download/v0.123.1/hugo_0.123.1_linux-amd64.deb
sudo apt install ./hugo....deb

Note: the apt package version is too old to use.

cd /path/to/hugo/blog/root/
sudo apt install -y snapd && sudo snap install hugo

hugo

encryption w/o modifing server #

1. to encrypt partial content (WARN: toc not encrypted!): hugo_encryptor https://github.com/Li4n0/hugo_encryptor/issues/22
2. to encrypt some entire pages: StatiCrypt

frequent config #

frequent cmd #

tip: for dropbox to ignore “public” dir.:

sudo apt-get install attr
attr -s com.dropbox.ignored -V 1 /path/to/Dropbox/.../hugo-blog/public/

2026-02-18 all in one watch.sh:

#!/bin/bash
# Initial build with clean
t_date_iso_like=$(date '+%Y-%m-%d_%H%M%S')
echo ============================== Started: ${t_date_iso_like}
echo "Cleaning and building..."
hugo --cleanDestinationDir && date +"%Y-%m-%d_%H%M%S" && python3 ./hugo_encryptor-master/hugo-encryptor.py
echo -n "============================== Init Done: "
date +'%Y-%m-%d_%H%M%S'

WATCH_DIRS="content/ layouts/ static/"
echo "Watching for changes in: $WATCH_DIRS"
fswatch $WATCH_DIRS | while read -r changed_file
do
    echo "Changed: $changed_file"
    date +"%Y-%m-%d_%H%M%S";
    echo "Rebuilding and encrypting..."
    hugo --cleanDestinationDir && date +"%Y-%m-%d_%H%M%S" && python3 ./hugo_encryptor-master/hugo-encryptor.py
    date +"%Y-%m-%d_%H%M%S"
    echo "Done. Waiting for next change..."
done

tip: for dropbox, need to delete Conflicts: vim rm_conflicts_and_hugo_build_and_sleep:

while true
do
  t_date_iso_like=$(date '+%Y-%m-%d_%H%M%S')
  echo ============================== Started: ${t_date_iso_like}
	find ./public/ -name '*Conflict*' | xargs  -d '\n' rm -r
	hugo
  date +'%Y-%m-%d_%H%M%S'
  python3 ./hugo_encryptor-master/hugo-encryptor.py
  echo -n "============================== Done: "
  date +'%Y-%m-%d_%H%M%S'
	sleep 0m 30s
done

or in one line: while true; do t_date_iso_like=$(date '+%Y-%m-%d_%H%M%S'); echo ============================== Started: ${t_date_iso_like}; find ./public/ -name '*Conflict*' | xargs -d '\n' rm -r; hugo; echo -n "============================== Done: "; date +'%Y-%m-%d_%H%M%S'; sleep 0m 30s; done

serving http #

hugo server #

(Only for dev env, with auto reload when .md file chagnes.)

nginx #

(略)

py ‘simple http server’ with port number 80: #

dated! use nginx !!!

cd public

sudo apt install -y python3-pip python3-setuptools libcap2-bin  && \
pip3 install SimpleHTTPServer SimpleHTTP404Server && \
ls /usr/bin/python3* | xargs -d '\n' sudo setcap 'cap_net_bind_service=+ep'

python3 -m SimpleHTTP404Server 80

themes #

learn & relearn: good: book-chapter-like sidebar; bad: no default TOC. loveIt: clean; bad: no default TOC.

add search for both cn & en in 5 min, by ttys3