Get Google search results, but without any ads, JavaScript, AMP links, cookies, or IP address tracking. Easily deployable in one click as a Docker app, and customizable with a single config file. Quick and simple to implement as a primary search engine replacement on both desktop and mobile.
Contents
- Features
- Install/Deploy Options
- Environment Variables and Configuration
- Usage
- Extra Steps
- Contributing
- FAQ
- Public Instances
- Screenshots
Features
- No ads or sponsored content
- No JavaScript*
- No cookies**
- No tracking/linking of your personal IP address***
- No AMP links
- No URL tracking tags (i.e. utm=%s)
- No referrer header
- Tor and HTTP/SOCKS proxy support
- Autocomplete/search suggestions
- POST request search and suggestion queries (when possible)
- View images at full res without site redirect (currently mobile only)
- Light/Dark/System theme modes (with support for custom CSS theming)
- Randomly generated User Agent
- Easy to install/deploy
- DDG-style bang (i.e.
!<tag> <query>
) searches - User-defined custom bangs
- Optional location-based searching (i.e. results near <city>)
- Optional NoJS mode to view search results in a separate window with JavaScript blocked
<sup>*No third party JavaScript. Whoogle can be used with JavaScript disabled, but if enabled, uses JavaScript for things like presenting search suggestions.</sup>
<sup>**No third party cookies. Whoogle uses server side cookies (sessions) to store non-sensitive configuration settings such as theme, language, etc. Just like with JavaScript, cookies can be disabled and not affect Whoogle's search functionality.</sup>
<sup>***If deployed to a remote server, or configured to send requests through a VPN, Tor, proxy, etc.</sup>
Install
There are a few different ways to begin using the app, depending on your preferences:
Heroku Quick Deploy
Provides:
- Easy Deployment of App
- A HTTPS url (https://<your app name>.herokuapp.com)
Notes:
- Requires a PAID Heroku Account.
- Sometimes has issues with auto-redirecting to
https
. Make sure to navigate to thehttps
version of your app before adding as a default search engine.
Render
Create an account on render.com and import the Whoogle repo with the following settings:
- Runtime:
Python 3
- Build Command:
pip install -r requirements.txt
- Run Command:
./run
Repl.it
Note: Requires a (free) Replit account
Provides:
- Free deployment of app
- Free HTTPS url (https://<app name>.<username>.repl.co)
- Supports custom domains
- Downtime after periods of inactivity (solution)
Fly.io
You will need a Fly.io account to deploy Whoogle. The free allowances are enough for personal use.
Install the CLI: https://fly.io/docs/hands-on/installing/
Deploy the app
flyctl auth login flyctl launch --image benbusby/whoogle-search:latest
The first deploy won't succeed because the default internal_port
is wrong.
To fix this, open the generated fly.toml
file, set services.internal_port
to 5000
and run flyctl launch
again.
Your app is now available at https://<app-name>.fly.dev
.
Koyeb
Use one of the following guides to install Whoogle on Koyeb:
- Using GitHub: https://www.koyeb.com/docs/quickstart/deploy-with-git
- Using Docker: https://www.koyeb.com/docs/quickstart/deploy-a-docker-application
pipx
Persistent install:
pipx install git+https://github.com/benbusby/whoogle-search.git
Sandboxed temporary instance:
pipx run --spec git+https://github.com/benbusby/whoogle-search.git whoogle-search
pip
pip install whoogle-search
$ whoogle-search --help usage: whoogle-search [-h] [--port <port number>] [--host <ip address>] [--debug] [--https-only] [--userpass <username:password>] [--proxyauth <username:password>] [--proxytype <socks4|socks5|http>] [--proxyloc <location:port>] Whoogle Search console runner optional arguments: -h, --help Show this help message and exit --port <port number> Specifies a port to run on (default 5000) --host <ip address> Specifies the host address to use (default 127.0.0.1) --debug Activates debug mode for the server (default False) --https-only Enforces HTTPS redirects for all requests --userpass <username:password> Sets a username/password basic auth combo (default None) --proxyauth <username:password> Sets a username/password for a HTTP/SOCKS proxy (default None) --proxytype <socks4|socks5|http> Sets a proxy type for all connections (default None) --proxyloc <location:port> Sets a proxy location for all connections (default None)
See the available environment variables for additional configuration.
Manual
Note: Content-Security-Policy
headers can be sent by Whoogle if you set WHOOGLE_CSP
.
Dependencies
- Python3
libcurl4-openssl-dev
andlibssl-dev
- macOS:
brew install openssl curl-openssl
- Ubuntu:
sudo apt-get install -y libcurl4-openssl-dev libssl-dev
- Arch:
pacman -S curl openssl
- macOS:
Install
Clone the repo and run the following commands to start the app in a local-only environment:
git clone https://github.com/benbusby/whoogle-search.git cd whoogle-search python3 -m venv venv source venv/bin/activate pip install -r requirements.txt ./run
See the available environment variables for additional configuration.
systemd Configuration
After building the virtual environment, you can add something like the following to /lib/systemd/system/whoogle.service
to set up a Whoogle Search systemd service:
[Unit] Description=Whoogle [Service] # Basic auth configuration, uncomment to enable #Environment=WHOOGLE_USER=<username> #Environment=WHOOGLE_PASS=<password> # Proxy configuration, uncomment to enable #Environment=WHOOGLE_PROXY_USER=<proxy username> #Environment=WHOOGLE_PROXY_PASS=<proxy password> #Environment=WHOOGLE_PROXY_TYPE=<proxy type (http|https|proxy4|proxy5) #Environment=WHOOGLE_PROXY_LOC=<proxy host/ip> # Site alternative configurations, uncomment to enable # Note: If not set, the feature will still be available # with default values. #Environment=WHOOGLE_ALT_TW=farside.link/nitter #Environment=WHOOGLE_ALT_YT=farside.link/invidious #Environment=WHOOGLE_ALT_RD=farside.link/libreddit #Environment=WHOOGLE_ALT_MD=farside.link/scribe #Environment=WHOOGLE_ALT_TL=farside.link/lingva #Environment=WHOOGLE_ALT_IMG=farside.link/rimgo #Environment=WHOOGLE_ALT_WIKI=farside.link/wikiless #Environment=WHOOGLE_ALT_IMDB=farside.link/libremdb #Environment=WHOOGLE_ALT_QUORA=farside.link/quetre # Load values from dotenv only #Environment=WHOOGLE_DOTENV=1 Type=simple User=<username> # If installed as a package, add: ExecStart=<python_install_dir>/python3 <whoogle_install_dir>/whoogle-search --host 127.0.0.1 --port 5000 # For example: # ExecStart=/usr/bin/python3 /home/my_username/.local/bin/whoogle-search --host 127.0.0.1 --port 5000 # Otherwise if running the app from source, add: ExecStart=<whoogle_repo_dir>/run # For example: # ExecStart=/var/www/whoogle-search/run WorkingDirectory=<whoogle_repo_dir> ExecReload=/bin/kill -HUP $MAINPID Restart=always RestartSec=3 SyslogIdentifier=whoogle [Install] WantedBy=multi-user.target
Then,
sudo systemctl daemon-reload
sudo systemctl enable whoogle
sudo systemctl start whoogle
Tor Configuration optional
If routing your request through Tor you will need to make the following adjustments. Due to the nature of interacting with Google through Tor we will need to be able to send signals to Tor and therefore authenticate with it.
There are two authentication methods, password and cookie. You will need to make changes to your torrc:
-
Cookie
-
Uncomment or add the following lines in your torrc:
ControlPort 9051
CookieAuthentication 1
DataDirectoryGroupReadable 1
CookieAuthFileGroupReadable 1
-
Make the tor auth cookie readable:
-
This is assuming that you are using a dedicated user to run whoogle. If you are using a different user replace
whoogle
with that user.
chmod tor:whoogle /var/lib/tor
chmod tor:whoogle /var/lib/tor/control_auth_cookie
-
-
Restart the tor service:
systemctl restart tor
-
Set the Tor environment variable to 1,
WHOOGLE_CONFIG_TOR
. Refer to the Environment Variables section for more details.- This may be added in the systemd unit file or env file
WHOOGLE_CONFIG_TOR=1
- This may be added in the systemd unit file or env file
-
-
Password
-
Run this command:
tor --hash-password {Your Password Here}
; put your password in place of{Your Password Here}
.- Keep the output of this command, you will be placing it in your torrc.
- Keep the password input of this command, you will be using it later.
-
Uncomment or add the following lines in your torrc:
ControlPort 9051
HashedControlPassword {Place output here}
; put the output of the previous command in place of{Place output here}
.
-
Now take the password from the first step and place it in the control.conf file within the whoogle working directory, ie. misc/tor/control.conf
- If you want to place your password file in a different location set this location with the
WHOOGLE_TOR_CONF
environment variable. Refer to the Environment Variables section for more details.
- If you want to place your password file in a different location set this location with the
-
Heavily restrict access to control.conf to only be readable by the user running whoogle:
chmod 400 control.conf
-
Finally set the Tor environment variable and use password variable to 1,
WHOOGLE_CONFIG_TOR
andWHOOGLE_TOR_USE_PASS
. Refer to the Environment Variables section for more details.- These may be added to the systemd unit file or env file:
WHOOGLE_CONFIG_TOR=1
WHOOGLE_TOR_USE_PASS=1
- These may be added to the systemd unit file or env file:
-
Manual (Docker)
- Ensure the Docker daemon is running, and is accessible by your user account
- To add user permissions, you can execute
sudo usermod -aG docker yourusername
- Running
docker ps
should return something besides an error. If you encounter an error saying the daemon isn't running, trysudo systemctl start docker
(Linux) or ensure the docker tool is running (Windows/macOS).
- Clone and deploy the docker app using a method below:
Docker CLI
Through Docker Hub:
docker pull benbusby/whoogle-search docker run --publish 5000:5000 --detach --name whoogle-search benbusby/whoogle-search:latest
or with docker-compose:
git clone https://github.com/benbusby/whoogle-search.git cd whoogle-search docker-compose up
or by building yourself:
git clone https://github.com/benbusby/whoogle-search.git cd whoogle-search docker build --tag whoogle-search:1.0 . docker run --publish 5000:5000 --detach --name whoogle-search whoogle-search:1.0
Optionally, you can also enable some of the following environment variables to further customize your instance:
docker run --publish 5000:5000 --detach --name whoogle-search \ -e WHOOGLE_USER=username \ -e WHOOGLE_PASS=password \ -e WHOOGLE_PROXY_USER=username \ -e WHOOGLE_PROXY_PASS=password \ -e WHOOGLE_PROXY_TYPE=socks5 \ -e WHOOGLE_PROXY_LOC=ip \ whoogle-search:1.0
And kill with: docker rm --force whoogle-search
Using Heroku CLI
heroku login heroku container:login git clone https://github.com/benbusby/whoogle-search.git cd whoogle-search heroku create heroku container:push web heroku container:release web heroku open
This series of commands can take a while, but once you run it once, you shouldn't have to run it again. The final command, heroku open
will launch a tab in your web browser, where you can test out Whoogle and even set it as your primary search engine.
You may also edit environment variables from your app’s Settings tab in the Heroku Dashboard.
Arch Linux & Arch-based Distributions
There is an AUR package available, as well as a pre-built and daily updated package available at
编辑推荐精选

酷表ChatExcel
大模型驱动的Excel数据处理工具
基于大模型交互的表格处理系统,允许用户通过对话方式完成数据整理和可视化分析。系统采用机器学习算法解析用户指令,自动执 行排序、公式计算和数据透视等操作,支持多种文件格式导入导出。数据处理响应速度保持在0.8秒以内,支持超过100万行数据的即时分析。


DeepEP
DeepSeek开源的专家并行通信优化框架
DeepEP是一个专为大规模分布式计算设计的通信库,重点解决专家并行模式中的通信瓶颈问题。 其核心架构采用分层拓扑感知技术,能够自动识别节点间物理连接关系,优化数据传输路径。通过实现动态路由选择与负载均衡机制,系统在千卡级计算集群中维持稳定的低延迟特性,同时兼容主流深度学习框架的通信接口。


DeepSeek
全球领先开源大模型,高效智能助手
DeepSeek是一家幻方量化创办的专注于通用人工智能的中国科技公司,主攻大模型研发与应用。DeepSeek-R1是开源的推理模型,擅长处理复杂任务且可免费商用。


问小白
DeepSeek R1 满血模型上线
问小白是一个基于 DeepSeek R1 模型的智能对话平台,专为用户提供高效、贴心的对话体验。实时在线,支持深度思考和联网搜索。免费不限次数,帮用户写作、创作、分析和规划,各种任务随时完成!


KnowS
AI医学搜索引擎 整合4000万+实时更新的全球医学文献
医学领域专用搜索引擎整合4000万+实时更新的全球医学文献,通过自主研发AI模型实现精准知识检索。系统每日更新指南、中英文文献及会议资料,搜索准确率较传统工具提升80%,同时将大模型幻觉率控制在8%以下。支持临床建议生成、文献深度解析、学术报告制作等全流程科研辅助,典型用户反馈显示每周可节省医疗工作者70%时间。


Windsurf Wave 3
Windsurf Editor推出第三次重大更新Wave 3
新增模型上下文协议支持与智能编辑功能。本次更新包含五项核心改进:支持接入MCP协议扩展工具生态,Tab键智能跳转提升编码效率,Turbo模式实现自动化终端操作,图片拖拽功能优化多模态交互,以及面向付费用户的个性化图标定制。系统同步集成DeepSeek、Gemini等新模型,并通过信用点数机制实现差异化的资源调配。


腾讯元宝
腾讯自研的混元大模型AI助手
腾讯元宝是腾讯基于自研的混 元大模型推出的一款多功能AI应用,旨在通过人工智能技术提升用户在写作、绘画、翻译、编程、搜索、阅读总结等多个领域的工作与生活效率。


Grok3
埃隆·马斯克旗下的人工智能公司 xAI 推出的第三代大规模语言模型
Grok3 是由埃隆·马斯克旗下的人工智能公司 xAI 推出的第三代大规模语言模型,常被马斯克称为“地球上最聪明的 AI”。它不仅是在前代产品 Grok 1 和 Grok 2 基础上的一次飞跃,还在多个关键技术上实现了创新突破。


OmniParser
帮助AI理解电脑屏幕 纯视觉GUI元素的自动化解析方案
开源工具通过计算机视觉技术实现图形界面元素的智能识别与结构化处理,支持自动化测试脚本生成和辅助功能开发。项目采用模块化设计,提供API接口与多种输出格式,适用于跨平台应用场景。核心算法优化了元素定位精度,在动态界面和复杂布局场景下保持稳定解析能力。


流畅阅读
AI网页翻译插件 双语阅读工具,还原母语级体验
流畅阅读是一款浏览器翻译插件,通过上下文智能分析提升翻译准确性,支持中英双语对照显示。集成多翻译引擎接口,允许用户自定义翻译规则和快捷键配置,操作数据全部存储在本地设备保障隐私安全。兼容Chrome、Edge、Firefox等主流浏览器,基于GPL-3.0开源协议开发,提供持续的功能迭代和社区支持。
推荐工具精选
AI云服务特惠
懂AI专属折扣关注微信公众号
最新AI工具、AI资讯
独家AI资源、AI项目落地

微信扫一扫关注公众号