Webvm

Virtual Machine for the Web

Generate Convert Improve

Install / Use

/learn @leaningtech/Webvm

About this skill

Quality Score

0/100

README

WebVM

This repository hosts the source code for https://webvm.io, a Linux virtual machine that runs in your browser.

Try out the new Alpine / Xorg / i3 graphical environment: https://webvm.io/alpine.html

WebVM is a server-less virtual environment running fully client-side in HTML5/WebAssembly. It's designed to be Linux ABI-compatible. It runs an unmodified Debian distribution including many native development toolchains.

WebVM is powered by the CheerpX virtualization engine, and enables safe, sandboxed client-side execution of x86 binaries on any browser. CheerpX includes an x86-to-WebAssembly JIT compiler, a virtual block-based file system, and a Linux syscall emulator.

Fork, deploy, customize
Running WebVM locally with a custom Debian mini disk image
Example customization: Python3 REPL
How to use Claude AI
Bugs and Issues
More links
Thanks to...
Versioning
License

Enable networking

Modern browsers do not provide APIs to directly use TCP or UDP. WebVM provides networking support by integrating with Tailscale, a VPN network that supports WebSockets as a transport layer.

Open the "Networking" panel from the side-bar
Click "Connect to Tailscale" from the panel
Log in to Tailscale (create an account if you don't have one)
Click "Connect" when prompted by Tailscale

WebVM now has access to machines in your own local Tailscale Network!

The world wide web

If you would like to access the public internet, you will need to set up an Exit Node on one of your non-WebVM tailscale network devices. See the "Advertise a device as an exit node" section of the Tailscale Exit Node quickstart guide for instructions. (The “Use an exit node” section can be skipped, as WebVM automatically uses an available exit node once one is advertised).

[!NOTE] While we support most network commands there are a few that rely on kernel-level features not available in modern browsers and are therefore not supported, most notably ping.You could use curl or wget for testing instead.

(Depending on your network speed, you may need to wait a few moments for the Tailscale Wasm module to be downloaded.) Once that is set up:

Log in with your Tailscale credentials.
Go back to the WebVM tab.
The Connect to Tailscale button in the Networking side-panel should be replaced by your IP address.

[!TIP] You can also check your connection status by checking the dot colour on the "connect to tailscale" button (which should now show your tailscale IP). On local network connectivity it will be orange, global will be green.

Using an authkey

As an alternative to manually logging in, you can add your tailscale auth Key at the end of the webvm URL.

https://webvm.io/#authKey=<your-key>

It is recommended to use an ephemeral key.

Selfhosting your tailscale network

We also support headscale, a selfhosted open source implementation of the Tailscale control server.

Though as headscale unfortunately doesn't support adding CORS headers. You will have to set up a proxy server to add them. Headscales instructions on doing so can be found here.

Once ready, add the following line to your location / block in your nginx config file.

 if ($http_origin = "https://webvm.io") {
            add_header 'Access-Control-Allow-Origin' "$http_origin";
			add_header 'Access-Control-Allow-Credentials' 'true' always;
        }

To log in to your headscale network add #controlUrl=<your-control-url> to the webVM url.

Notes:

If self hosting, replace "https://webvm.io" with your own url.
This is equivelant to the tailscale --login-server command line option.
If used with authkey, don't forget to seperate the URL fragments with a & inbetween.

Fork, deploy, customize

Fork the repository.
Enable Github pages in settings.
- Click on Settings.
- Go to the Pages section.
- Select Github Actions as the source.
  - If you are using a custom domain, ensure Enforce HTTPS is enabled.
Run the workflow.
- Click on Actions.
- Accept the prompt. This is required only once to enable Actions for your fork.
- Click on the workflow named Deploy.
- Click Run workflow and then once more Run workflow in the menu.
After a few seconds a new Deploy workflow will start, click on it to see details.
After the workflow completes, which takes a few minutes, it will show the URL below the deploy_to_github_pages job.

You can now customize dockerfiles/debian_mini to suit your needs, or make a new Dockerfile from scratch. Use the Path to Dockerfile workflow parameter to select it.

If you would like to use our full desktop Alpine image, you can find it's dockerfile here.
For more information on creating custom images, see our Custom disk Image documentation.

Run WebVM locally with a custom Debian mini disk image

1. Clone the WebVM Repository

git clone https://github.com/leaningtech/webvm.git
cd webvm

2. Download the Debian mini Ext2 image

Run the following command to download the Debian mini Ext2 image:

wget "https://github.com/leaningtech/webvm/releases/download/ext2_image/debian_mini_20230519_5022088024.ext2"

(You can also build your own disk image by selecting the "Upload GitHub release" workflow option)

3. Update the configuration file

Edit `config_public_terminal.js` to reference your local disk image:

Replace:

"wss://disks.webvm.io/debian_large_20230522_5044875331.ext2"

With:

"/disk-images/debian_mini_20230519_5022088024.ext2"

(Use an absolute or relative URL pointing to the disk image location.)
Replace "cloud" with the correct disk image type: "bytes"

4. Build WebVM

Run the following commands to install dependencies and build WebVM:

npm install
npm run build

The output will be placed in the build directory.

5. Configure Nginx

Create a directory for the disk image:

mkdir disk-images
mv debian_mini_20230519_5022088024.ext2 disk-images/

Modify your nginx.conf file to serve the disk image. Add the following location block:

location /disk-images/ {
	root .;
	autoindex on;
}

6, Start Nginx

Run the following command to start Nginx:

nginx -p . -c nginx.conf

Nginx will automatically serve the build directory.

7. Access WebVM

Open a browser and visit: http://127.0.0.1:8081.

Enjoy your local WebVM!

Example customization: Python3 REPL

The Deploy workflow takes into account the CMD specified in the Dockerfile. To build a REPL you can simply apply this patch and deploy.

diff --git a/dockerfiles/debian_mini b/dockerfiles/debian_mini
index 2878332..1f3103a 100644
--- a/dockerfiles/debian_mini
+++ b/dockerfiles/debian_mini
@@ -15,4 +15,4 @@ WORKDIR /home/user/
 # We set env, as this gets extracted by Webvm. This is optional.
 ENV HOME="/home/user" TERM="xterm" USER="user" SHELL="/bin/bash" EDITOR="vim" LANG="en_US.UTF-8" LC_ALL="C"
 RUN echo 'root:password' | chpasswd
-CMD [ "/bin/bash" ]
+CMD [ "/usr/bin/python3" ]

How to use Claude AI

To access Claude AI, you need an API key. Follow these steps to get started:

1. Create an account

Visit Anthropic Console and sign up with your e-mail. You'll receive a sign in link to the Anthropic Console.

2. Get your API key

Once logged in, navigate to Get API keys.
Purchase the amount of credits you need. After completing the purchase, you'll be able to generate the key through the API console.

3. Log in with your API key

Navigate to your WebVM and hover over the robot icon. This will show the Claude AI Integration tab. For added convenience, you can click the pin button in the top right corner to keep the tab in place.
You'll see a prompt where you can insert your Claude API key.
Insert your key and press enter.

4. Start using Claude AI

Once your API key is entered, you can begin interacting with Claude AI by asking questions such as:

"Solve the CTF challenge at /home/user/chall1.bin. Note that the binary reads from stdin."

Important: Your API key is private and should never be shared. We do not have access to your key, which is not only stored locally in your browser.

Bugs and Issues

Please use Issues to report any bug. Or come to say hello / share your feedback on Discord.