this kind of dismissive comments is why many apps have an awful usability. If someone thinks a web interface would be easier than a text terminal, there's at least one customer waiting for a product (which either doesn't exist or they could not googlify)
I hired a programmer and after giving him his Linux laptop let him set up a few things. A couple hours later he asked me where he could get PuTTY for it, and I recognized a huge gap in my interview coverage.
> Puttygen software is not created, nor supported by Puttygen.com. The program has been tested and is believed to be safe. [...] The use of Puttygen through Puttygen.com is done at your own discretion and risk
Plan9 is funny because it's what UNIX might look like if the people working on UNIX understood UNIX, i.e. everything is a file and simple primitives are composed into complex systems.
They had the benefit of hindsight and bigger hardware, but UNIX got too popular and now we're struggling to move past it. It would have been interesting to see what the fourth try would be like (though looking at Go I would probably not completely like it).
Funny enough, that right there is the actual fundamental problem here.
I am reminded of a post or blog long ago that talked about programmable thermostats and how awful they are for most people to use despite how powerfully in the weeds one can get with them. Basically summarizing the issue as something like “People do not want to learn your arcane system, they just want the benefit it’s advertising”. A good UI knows how to minimize that gap.
No. It’s just that now more people are using Linux the more the ux decisions that were made 40 years ago will be questioned.
Almost all dev facing machines have ssh server installed and accessible.
Why ssh terminal has to look like character-only trash from 1960s? Why a TUI is the best thing we pipe through ssh? Why I cannot watch a 4k movie in the terminal or browse the web using pinch to zoom ?
On various Mozilla forums that I saw, the discussion was basically:
1. We can't just allow the browser to connect to any socket, since many either explicitly don't want browsers connecting to them, or are oblivious to browsers.
2. ...so we need to also add some sort of allow list
3. ...this is getting too complicated for such a niche feature.
So I think the nicheness was the high-order bit here.
Everything they mention as "missing", or "novel" has been part of Cockpit for over a decade, from socket-based web server connection, backend-frontend separation for server apps and the whole idea of a server console with shell access itself.
To answer them: "Isn’t it weird that this doesn’t already exist?" - No, it's not, because it has existed for ages.
I don't think I've met many pairs of people I could ask "on a scale of 1 to 10, how biased/loaded do you think ${example} is" and get told the same exact number by both for the majority of the examples given.
Now apply that to the n people reading a given post or comment! If those commenters try to communicate on what they think is "fair game" for the given conversation, then two comments deep in and you might already be at a 7 when the author thought they were at a 3. In more extreme cases, two people could misunderstand each other through text and simply go from a 1 to a 7 in a single comment, spending the rest of the time shooting back loaded replies at each other instead of thinking about the topic together.
It's a very human reaction we all tend towards, even when we set out our intents to do the "always reply with..." mindset instead of a tit-for-tat one. That's why I always start with the ideal approach - I can count on myself to help foul it up :D.
Thanks for pointing this out. I'm not hating on Cockpit, but Outer Loop (with Outer Shell) has solved a lot more of the stack. Cockpit accepts the constraints of living in existing browsers, so it requires exposing a port to the internet or using some SSH port forwarding tool. Whereas I built a dedicated browser to push capabilities so that users can get a "Just point me to a server" flow.
This thread has been useful -- I think Cockpit will also work great in Outer Loop. And it will be easy to add it as an app in Outer Shell.
> it requires exposing a port to the internet or using some SSH port forwarding tool
This sentence is bizarre to me. Your SSH-based solution also requires exposing a port to the internet and installing a special tool (on both server and client!). What's so special about SSH that using HTTPS is a problem but using SSH isn't?
The industry also tried the whole "use the web browser to run native binaries" thing with ActiveX (and the unity web player I guess). The idea was thrown out along with flash and java applets for what I presume were security and portability reasons.
If you can SSH to a machine, you can use Outer Loop and Outer Shell, without having to do any sudo commands or expose anything new to the network. The browser + SSH client combined into a single app leads to nice user experiences like this. The final section of the post was saying that it's strange such a thing doesn't exist already.
> without having to do any sudo commands or expose anything new to the network.
Again I'm not understanding the distinction. I don't need to run sudo commands to install a web server, and depending on your definition of "exposing something new" to the network then either I don't have to do that either or your solution also does that.
Something is getting downloaded and run on the remote machine, correct? Why is it problematic for that something to be a web server (with SSH-forwarding I guess) instead of this custom thing?
And why install anything on the server at all if it'll just serve a binary that downloads and runs on your local computer anyway? For example, if I type `sftp://username@server.domain/file/path` into my file manager's address bar, I get the nice file browsing experience you demonstrate without installing anything on my computer or the server.
EDIT: OK, after reading through your earlier posts, I think the value proposition really is just that you've implemented a slightly better UX for proxying remote web servers via ssh, and that the "run native code" thing is an independent idea you are also pursuing. So the answer to the question "isn't this just proxying an http server over ssh" is basically yes.
I think I incorrectly read this as attempting to propose a radically new idea and not as an incremental improvement to the status quo.
> it requires exposing a port to the internet or using some SSH port forwarding tool
I think what they meant is that the SSH server can be behind your webserver and not have to have its own public IP exposed directly... but of course there are an abundance of proxy-related solutions already.
Cockpit has a "remote" host connection feature solving this exact pain-point - "Just point me to a server":
You install the Cockpit web service on one host (along with its backend and extensions), and on other hosts you may have - install only the backend of the stack (4-7 packages available via deb backports & other dist repos). The web front host is then able to access any other machine via ssh (if keys and policies permit that) and display info or manage that host. All ports aside from the web front and ssh between your hosts remain as is. It is a decentralized design.
To the author's defense: Cockpit is Linux only, and they seem to intend on making this also available on Windows and Mac.
Still, I don't see the appeal they seem to do, especially since it relies so much on SSH. The biggest use case I can think for something like this in the real world is something like first-time setup or MDM, and on both situations setting up SSH to begin with has the same level of friction they're trying to remove.
Windows has quite a lot of remote admin tools that work pretty transparently over the network though.
The issue is that they're historically never turned on or heavily restricted.
Where the user is involved though RDP is a world class remote desktop never exceeded by Linux anywhere.
If someone wants to impress me, point Claude at Wayland and get it so I can seamlessly open remote RDP from somewhere else, lock the local user session and resume it on the remote desktop, then walk back to the original terminal and continue working in that same user session. This worked perfectly over 20 years ago.
i'm trying to understand how outer shell works here. on the website you give the following as your motivation:
> Apps like Jupyter and Tensorboard are not typically visible to standard web browsers if they’re running on remote servers, because it would be terribly unsafe to let the whole internet touch this app. Instead, they run on a local port on the server, which your computer can’t access directly.
> Classically, to get access to these, you had to open a new terminal and run:
is this true? isn't the normal thing just to do this ssh forwarding for prototyping, then for deployment, you set up a website like myjupyternotebook.com, and then set up auth so that others can't access it. HTTP basic auth is not too much work.
if you want SSH, not HTTP, to be what's publicly exposed, there's other options too, like putting it behind a VPN or tunnel.
all this to say, outer loop is super cool, but I don't get it. I must be missing something about why you built it, so could you help me understand?
I think there are different clusters of people who use servers, SSH, etc.
I'm closer to the cluster that uses them for deep learning experiments, GPU kernel optimization, robot development (a robot is just a server that moves!)... use cases where you are explicitly using a remote computer.
For this cluster of people, I think this tool feels more intuitive than the flow you suggest. But maybe I'm projecting!
And, to me, this just feels like one of the fundamental things that could exist; it's like a graphical operating system, but remote-first.
I guess it saves you the hassle of dealing with reverse proxies and TLS certs if your use case is "userbase is 1 person and it is me, and i only access services from a desktop os"
Btw, if you find yourself sending a lot of ports over ssh, you can also consider the option of having ssh start a socks5 proxy
ssh -D 4711 -q -C -N user@host
sets localhost:4711 up as a socks5 proxy you can tell your browser to use
...
A wireguard VPN is better of course; among other things because ssh is multiplexing over a single TCP connection and will encounter head of line blocking (where one dropped packet blocks all forwarded traffic until resent)
I also did some experiments some time ago. The thing this is missing for me is the ability to also run arbitrary commands other that just using a few premade apps. In fact I think this stuff becomes really interesting when you put a real "shell" on top of this.
And I don't mean a classical posix shell, something that can be used to leverage the full power of the custom ui and frontend. Also a must have is "nestable connections".
The experiment I was doing was with a web interface and a statically compiled Go backend (for easy deployment via ssh). Maybe some day I will finish it xD
Do not do this. There are many, many excellent long-standing security and "web control plane isolation" reasons browsers are not permitted generic socket permissions.
The closest mechanical analog that comes to mind is why 3-wheeled ATVs are a bad idea.
- sockets are blocked by default, until they are added to an allow-list explicitly on the server side
- True sudo awareness ensures root sockets aren't reachable without the sudo password. (This capability is important, because otherwise you create an incentive for people to run root backends with user-accessible sockets.)
Lovely writeup! I'll bookmark this for my own research.
My terminal's "clickity clackity" features [0] are local to the machine so I lose graphical-ness as soon as we remote in somewhere.
That's starting to change a bit with offline replay [1] where the native GUI and TUI work in tandem to unlock some rewind. But there's quite a road ahead and I love seeing others experiment properly. (Terminals are massively underserved.)
I think the approach here where interfacing with a device is considered from first principles is one that is rarely taken on, and this is a thought provoking implementation. Kudos.
This reminds me of an idea that I build a PoC of many years ago (maybe 2013 if I recall) that I always felt was the nugget of a useful idea. You would SSH into a server and processes on the other end would emit data which was then displayed in a webapp that was served from a localhost port, with a local backend that consumed the data. So for example a short-lived web-based remote 'top'. I did it as part of a company-internal hackathon and thought it was really cool, but nobody else was impressed with it. It was a very half-baked idea, and this looks like a fully-baked version of it. I'll check it out.
That's interesting idea. If we put into CLI with some ANSI escape code, that may become something real. Imagine a normal terminal app just render part of the UI in web and communicating in UNIX socket. While doing the fancy UI, everything is still controllable with keyboard, and optionally with mouse. The UI will fallback to text UI for older terminal
If your UI is not fully controllable with a keyboard, the same forces that made that happen will eventually make a mouse mandatory for this hypothetical tech stack too.
The terminal has no Platonic quality of being keyboard only. It is an accident of history and the limitations it has had. Remove the limitations and remove the accident of history and you will just end up drawn into the strange attractor of GUIs, warts and all.
There could be a brief honeymoon where the tech stack looks like some of you are imagining in your heads, but it would only last as long as it wasn't used by very many people. Google "gemini protocol" for a similar situation. That protocol has basically a cap on how popular it could possibly get before it just turned into HTTP B as the rest of the world forcibly upgraded it regardless of what the core project thinks. They exist in the shadow of HTTP, as the terminal exists in the shadow of GUIs. This is not a bad thing. It is what lets them be what they are. The shadows of GUIs or HTTP is large and there is plenty of space to be. Trying to give the terminal more GUI capabilities is like trying to give Gemini more web capabilities; you'll just end up in the same place, only with less refinement.
No no not something on top of the UI stack. They also need framebuffer support so they are big headache to setup on headless server.
What I mean is that we can bring some web tech to terminal natively. We don't even need a separated shell. Security and bi-directional communication is built by default because of UNIX socket. But we still need to think how to handle stuff like cookie, local storage, external CSS / JS, ...
Web technologies are significantly larger and more complex than framebuffers, and they don't even let you start arbitrary programs like Chrome under them.
Feedback: Home pages of each of Outer Loop, Outer Frame and Outer Shell contain basic intro of each instead of a link redirecting to them. By the time I click the link and on the new Outer X I have already what Outer X I came from and what it meant.
One of the more interesting pieces of Microsoft software is the Windows Admin Center where it's a web app to configure a Windows Server. Ideally, it was made for core installs where there's no GUI but it's there as a viable web management panel.
The tool from OP and WAC are pretty similar in terms of functionality and usecase. Why would you want this? Well, imagine your team needing to be able to do server functions but you have less technical team members to do it for you, which is very often the case in big places, most people are familiar with the web browser and having a website to do these sorts of actions makes it easier to have things done in one place without a lot of tools like Remote Desktop, SSH, WinRM, etc. configured.
At the risk of being considered a snob I don’t want someone who can’t deal with SSH or RDP configuring servers within my company. If you can’t work out how to SSH into the server you sure as hell aren’t going to work out how to safely expose network services on it.
Within your company, sure. But there's some engineers (think medical) who know standards like DICOM and PACS imaging but aren't familiar at all with OS internals or systems administration.
As someone managing various servers, both at home and at work, I see how this can be really useful. I see it not in the production space yet but rather in the experimenting, using a Linux machine as a second compute device!
So regarding your last point, I'm convinced. I think it is useful!
The one fact that is bugging me is that now it requires a client specific app, with GUI, on my PC and I wonder if using ssh port forwarding could reduce the surface. I mean I wonder if either having a rich client that executes commands via ssh or a rich server (including Web Server) with ssh port wouldn't suffice, so that I can avoid installing stuff on the server AND on my computer.
Just had a quick look but I like the look so far. I’ve been thinking along similar lines for ages but never quite got around to making something. I very much support any effort to make remoting less dependent on the archaic character grid.
I’m good with just tailscale and self-hosted web-apps. Seems the main selling point is either native UX or reduced barriers to entry security-wise. I like barriers to entry.
I can’t make up my mind if I love it or hate it. On one hand this is like SSHapi on the other there’s no structure, no contract… i had similar doubts with Cockpit.
It's a cool video and I like the idea in general. The author mentions that the code runs in a sandbox. I'm surprised that WASM hasn't come up. You want the code to be platform agnostic anyway (it should run whether you start Outshell on Linux, macOS or whatever on different CPU architectures).
Being able to initiate a shell app from a regular remote ssh CLI prompt (like "ApacheConfig myhost.com" or "Editor ~/myrepo") might improve integration with people's existing CLI workflows.
It does need an agent that starts with every X or Wayland session and waits for requests from remote SSH sessions to start an app.
I don't really know what outerframe frame is. I tried to understand from the video and the blog but I'm still not sure what it is. Is it like a web browser but instead of DOM, HTML and JS you have Swift and SwiftUI running in a sandbox?
If so how would that work on non Apple devices? Also how much will that sandbox protect you?
It's a fun heretical idea, moving away from a "cross-platform" web to a "multi-platform" web. It's a cross-platform protocol that hands off to platform-specific frontend code. I think it's a natural direction for the web, in a world where LLMs can translate to other platforms.
I thought this looks interesting, but was a little confused with what appears to be MacOS-only support at https://outerloop.sh/? I'm running Ubuntu 24.04, I kind of assumed from context that it'd be something I could spin up in a few minutes just to give it a go?
Also worth noting, my decision to give it a go relied mostly on the fact that I couldn't quite work out what the product is. Having "Outer Shell" and "Outer Loop" described as distinct-but-connected entities is a little confusing, IMO, which do I need to install, on what, and in what order?
I have also been having trouble grasping the difference between Outer Loop and Outer Shell. I thought maybe one was the desktop browser app for macOS and the other was something running locally on the Pi to create the socket. However, after bouncing between the links for the two, I don't think that assumption was correct.
I am not sure I'd use this over exposing websites with wireguard as those will automatically work across platforms. But it looks like you could create some really cool experiences with it, and I'm happy people are exploring this space.
I wrote an early version of the Cylance AV desktop client. The UI side was a web app that talked to its windows service backend using HTTP over windows pipes. This was surprisingly easy to do using WCF.
I'm confused -- does this compile it live when the server ships code?
How do we resolve dependencies, toolset etc..
Is the idea to just pick an old enough platform toolchain you expect to be present?
In all cases, the code is pre-compiled. A user never waits for anything to compile. When Outer Loop installs Outer Shell, it downloads pre-compiled binaries to the server. For Linux these are compiled against a manylinux ABI. Ditto for when Outer Shell installs one of the bundled apps. When a backend serves a native "web" app over HTTP it sends already-compiled ARM (or x86) code to the client.
Dependencies are less of a concern for the frontend binaries. For backends, I use a dependency-light approach, static-linking anything that's needed. Of course, people are welcome to do backends however they want, and just tell Outer Shell about the systemd/launchd units via the API. I used this no-dependency approach to keep everything lightweight and to keep install steps trivial, but admittedly it pushes me in certain directions (for example, using custom binary formats rather than sqlite).
I'm actually way more interested in option 2 - the VNC-like experience.
TUI apps are convenient over SSH because they're right there in your terminal. But they suck because they're restricted to shitty monospaced character grids. Why can't we have something more like VNC over SSH? Like, `top` and `micro` but with good graphics?
I did try doing something like that with the Kitty graphics protocol and you can get kind of close..ish, but it's really restricted by having to send everything as PNGs.
Anyway upvote for not being blinkered and thinking terminals are just for CLI stuff and must be forever.
In general I would like to see a web browser escape sequence for console applications. Just send a command to the terminal to connect a web browser to your stdin/out and present any UI you want over html. The terminal can then open a regular socket listening on localhost and act as a CGI server. For security the terminal should pick a random IP in the localhost range and a random URL. Technically that is security by obscurity, but guessing a cryptographically secure URL should be hard enough for attackers. The reasons to do it as an escape sequence and not just have the application open a socket and start the browser are: To enable remote GUI; To avoid the complexity of each application implementing networking; To enable better desktop integration, since the terminal itself is part of the Desktop Environment, so it can start a DE-specific browser, preferably in single-application mode. Also, it should be possible to automatically put the application in the background so you basically just run GUI applications like normal.
I don't have a dog in this fight, and anyway dogfighting is bad, but the intro to the Wikipedia article[0] reads:
> An operating system shell is a computer program that provides relatively broad and direct access to the system on which it runs. The term shell refers to how it is a relatively thin layer around an operating system.
> Most shells are command-line interface (CLI) programs. Some graphical user interfaces (GUI) also include shells.
The last line I think supports the notion that the term "shell" at least implies a CLI, but I can understand both positions.
The earliest versions of MacOS, all the way up through 9, had a ROM call at 0xA9F4 which was labeled `_exitToShell`. In the days before pre-emptive multitasking, this instruction's job was to force the current application to close and return the user to the MacOS desktop (the Finder). The "shell" in this context being the desktop user interface.
I wondered if this would be controversial. It all depends where you grew up.
> Cairo, like Chicago, had a new shell (Microsoft’s favorite word for the user interface for launching programs and managing files) and a new file system
Not really no. I’ve been using shells and authoring new ones for around 40 years across a variety of platforms. The term has always been pretty loosely defined because as technology evolved the term “shell” was borrowed. So like I said, a shell can refer to a graphical core just as much as a text-based one. You can get web shells too.
The original intent was that a shell is a thin wrapper on top of the OS to expose the hosts capabilities. But that hasn’t been an apt description for most of those 40 years.
command line shell vs graphical shell. My first experience with a graphical shell was dosshell[1]. For a while we called the Windows 3.1 interface "the shell". I guess the terminology has changed since that time.
111 comments:
This appears to me like a solution in search of a problem, like many others before it...the quote below seems relevant to this effort.
"Those who do not understand Unix are condemned to reinvent it, poorly." ~Henry Spencer
this kind of dismissive comments is why many apps have an awful usability. If someone thinks a web interface would be easier than a text terminal, there's at least one customer waiting for a product (which either doesn't exist or they could not googlify)
I hired a programmer and after giving him his Linux laptop let him set up a few things. A couple hours later he asked me where he could get PuTTY for it, and I recognized a huge gap in my interview coverage.
Indeed, a new hire should be able to use Google to find https://puttygen.com/download-putty#Download_PuTTY_on_Linux_... in short order
evil grin
https://www.chiark.greenend.org.uk/~sgtatham/putty/
puttygen.com looks super fishy, the disclaimer:
> Puttygen software is not created, nor supported by Puttygen.com. The program has been tested and is believed to be safe. [...] The use of Puttygen through Puttygen.com is done at your own discretion and risk
Edit: or is that the evil grin?
multiple issues right
I dislike this story and it’s because I can believe it.
Gotta auto-reject anyone listing Windows experience on resume
works for me. are you hiring? :-)
that seems a little harsh. I think there is a real usability gap which this takes a crack at.
Some ideas like using viewing a linux dir over _ssh_ using native UI components.. seem cool.
I do agree, some of these do seem like they have already been solved in other ways (like an sshfs mount).
This resembles Plan9 more than UNIX. I wouldn't put UNIX up on a pedestal.
Plan9 is funny because it's what UNIX might look like if the people working on UNIX understood UNIX, i.e. everything is a file and simple primitives are composed into complex systems.
They had the benefit of hindsight and bigger hardware, but UNIX got too popular and now we're struggling to move past it. It would have been interesting to see what the fourth try would be like (though looking at Go I would probably not completely like it).
yea, as sibling said. p9 was not possible on pdp11. what was possible there was .. v7 and 2bsd. see https://github.com/felipenlunkes/run-ancient-unix
p9 was done when "current state of unix" was already fixed in form of aix, sysv and bsds, it suffered the same fate as say beos.
> "Those who do not understand Unix
Funny enough, that right there is the actual fundamental problem here.
I am reminded of a post or blog long ago that talked about programmable thermostats and how awful they are for most people to use despite how powerfully in the weeds one can get with them. Basically summarizing the issue as something like “People do not want to learn your arcane system, they just want the benefit it’s advertising”. A good UI knows how to minimize that gap.
I mean that's true but the number of UIs which simply don't add access to necessary features in the name of "simplicity" is enormous.
The poster child of this is the Microsoft Office ribbon.
No. It’s just that now more people are using Linux the more the ux decisions that were made 40 years ago will be questioned.
Almost all dev facing machines have ssh server installed and accessible.
Why ssh terminal has to look like character-only trash from 1960s? Why a TUI is the best thing we pipe through ssh? Why I cannot watch a 4k movie in the terminal or browse the web using pinch to zoom ?
> Why a TUI is the best thing we pipe through ssh?
`ssh -XC` (look up SSH X forwarding). You can also easily tunnel remote desktop over ssh.
> Why I cannot watch a 4k movie in the terminal or browse the web using pinch to zoom ?
Kitty, sixel, and iterm2
I think this is a `There’s no such thing as bad publicity`
I like the idea of separating the frontend and backend of a graphical app. But I feel like this is hardly a novel idea, maybe I'm missing something.
I take it you don't know about "X11Forwarding yes" or "html5 web app"
That is a security concern, that's why it isn't implemented. At least raw unix socks. You can have WebSockets and other ports only limited to http.Quick response regarding security:
On various Mozilla forums that I saw, the discussion was basically: 1. We can't just allow the browser to connect to any socket, since many either explicitly don't want browsers connecting to them, or are oblivious to browsers. 2. ...so we need to also add some sort of allow list 3. ...this is getting too complicated for such a niche feature.
So I think the nicheness was the high-order bit here.
(FYI, Outer Loop does add an allow-list: https://outerloop.sh/unix-domain-sockets/)
Author apparently has never heard about Cockpit.
Everything they mention as "missing", or "novel" has been part of Cockpit for over a decade, from socket-based web server connection, backend-frontend separation for server apps and the whole idea of a server console with shell access itself.
To answer them: "Isn’t it weird that this doesn’t already exist?" - No, it's not, because it has existed for ages.
> Be kind. Don't be snarky. Converse curiously; don't cross-examine. Edit out swipes.
Sincerely, HN Guidelines Police :-)
https://news.ycombinator.com/newsguidelines.html
I get it, but if the author of the article uses a biased and loaded language, I think it's fair game to do the same in the comments.
I don't think I've met many pairs of people I could ask "on a scale of 1 to 10, how biased/loaded do you think ${example} is" and get told the same exact number by both for the majority of the examples given.
Now apply that to the n people reading a given post or comment! If those commenters try to communicate on what they think is "fair game" for the given conversation, then two comments deep in and you might already be at a 7 when the author thought they were at a 3. In more extreme cases, two people could misunderstand each other through text and simply go from a 1 to a 7 in a single comment, spending the rest of the time shooting back loaded replies at each other instead of thinking about the topic together.
It's a very human reaction we all tend towards, even when we set out our intents to do the "always reply with..." mindset instead of a tit-for-tat one. That's why I always start with the ideal approach - I can count on myself to help foul it up :D.
I don't believe in that kind of response. Anything that one can say in rage or anger can be communicated in a calm and measured response.
If I'm not mistaken cockpit is web UI and doesn't run native code, important differences.
Thanks for pointing this out. I'm not hating on Cockpit, but Outer Loop (with Outer Shell) has solved a lot more of the stack. Cockpit accepts the constraints of living in existing browsers, so it requires exposing a port to the internet or using some SSH port forwarding tool. Whereas I built a dedicated browser to push capabilities so that users can get a "Just point me to a server" flow.
This thread has been useful -- I think Cockpit will also work great in Outer Loop. And it will be easy to add it as an app in Outer Shell.
> it requires exposing a port to the internet or using some SSH port forwarding tool
This sentence is bizarre to me. Your SSH-based solution also requires exposing a port to the internet and installing a special tool (on both server and client!). What's so special about SSH that using HTTPS is a problem but using SSH isn't?
The industry also tried the whole "use the web browser to run native binaries" thing with ActiveX (and the unity web player I guess). The idea was thrown out along with flash and java applets for what I presume were security and portability reasons.
If you can SSH to a machine, you can use Outer Loop and Outer Shell, without having to do any sudo commands or expose anything new to the network. The browser + SSH client combined into a single app leads to nice user experiences like this. The final section of the post was saying that it's strange such a thing doesn't exist already.
FYI I made the same ActiveX connection here in the closing of the FAQ in the previous blog post about this native platform: https://probablymarcus.com/blocks/2026/05/10/like-a-web-view... I'm particularly proud of that paragraph.
> without having to do any sudo commands or expose anything new to the network.
Again I'm not understanding the distinction. I don't need to run sudo commands to install a web server, and depending on your definition of "exposing something new" to the network then either I don't have to do that either or your solution also does that.
Something is getting downloaded and run on the remote machine, correct? Why is it problematic for that something to be a web server (with SSH-forwarding I guess) instead of this custom thing?
And why install anything on the server at all if it'll just serve a binary that downloads and runs on your local computer anyway? For example, if I type `sftp://username@server.domain/file/path` into my file manager's address bar, I get the nice file browsing experience you demonstrate without installing anything on my computer or the server.
EDIT: OK, after reading through your earlier posts, I think the value proposition really is just that you've implemented a slightly better UX for proxying remote web servers via ssh, and that the "run native code" thing is an independent idea you are also pursuing. So the answer to the question "isn't this just proxying an http server over ssh" is basically yes.
I think I incorrectly read this as attempting to propose a radically new idea and not as an incremental improvement to the status quo.
> it requires exposing a port to the internet or using some SSH port forwarding tool
I think what they meant is that the SSH server can be behind your webserver and not have to have its own public IP exposed directly... but of course there are an abundance of proxy-related solutions already.
Cockpit has a "remote" host connection feature solving this exact pain-point - "Just point me to a server": You install the Cockpit web service on one host (along with its backend and extensions), and on other hosts you may have - install only the backend of the stack (4-7 packages available via deb backports & other dist repos). The web front host is then able to access any other machine via ssh (if keys and policies permit that) and display info or manage that host. All ports aside from the web front and ssh between your hosts remain as is. It is a decentralized design.
It's a very, very thin web layer on top of native code:
https://cockpit-project.org/guide/latest/features.html
To the author's defense: Cockpit is Linux only, and they seem to intend on making this also available on Windows and Mac.
Still, I don't see the appeal they seem to do, especially since it relies so much on SSH. The biggest use case I can think for something like this in the real world is something like first-time setup or MDM, and on both situations setting up SSH to begin with has the same level of friction they're trying to remove.
Windows has quite a lot of remote admin tools that work pretty transparently over the network though.
The issue is that they're historically never turned on or heavily restricted.
Where the user is involved though RDP is a world class remote desktop never exceeded by Linux anywhere.
If someone wants to impress me, point Claude at Wayland and get it so I can seamlessly open remote RDP from somewhere else, lock the local user session and resume it on the remote desktop, then walk back to the original terminal and continue working in that same user session. This worked perfectly over 20 years ago.
I never heard of cockpit either
what is it?
https://cockpit-project.org/
i'm trying to understand how outer shell works here. on the website you give the following as your motivation:
> Apps like Jupyter and Tensorboard are not typically visible to standard web browsers if they’re running on remote servers, because it would be terribly unsafe to let the whole internet touch this app. Instead, they run on a local port on the server, which your computer can’t access directly.
> Classically, to get access to these, you had to open a new terminal and run:
> ssh -L 24601:localhost:8889 mrcslws@lambda4.mycompany.com &
> ssh -L 24602:localhost:6006 mrcslws@lambda4.mycompany.com &
is this true? isn't the normal thing just to do this ssh forwarding for prototyping, then for deployment, you set up a website like myjupyternotebook.com, and then set up auth so that others can't access it. HTTP basic auth is not too much work.
if you want SSH, not HTTP, to be what's publicly exposed, there's other options too, like putting it behind a VPN or tunnel.
all this to say, outer loop is super cool, but I don't get it. I must be missing something about why you built it, so could you help me understand?
I think there are different clusters of people who use servers, SSH, etc.
I'm closer to the cluster that uses them for deep learning experiments, GPU kernel optimization, robot development (a robot is just a server that moves!)... use cases where you are explicitly using a remote computer.
For this cluster of people, I think this tool feels more intuitive than the flow you suggest. But maybe I'm projecting!
And, to me, this just feels like one of the fundamental things that could exist; it's like a graphical operating system, but remote-first.
I still don't get it. Isn't this what X11 forwarding is for?
It's too slow. I mention this in the video at 1:20 - 1:50.
I guess it saves you the hassle of dealing with reverse proxies and TLS certs if your use case is "userbase is 1 person and it is me, and i only access services from a desktop os"
Ever since I started using Caddy, doing that has been soooo easy.
Download the binary, make a Caddyfile
And then just "./caddy start"I just use https://tuns.sh which has a handy bash script to make the ssh tunnel simple
Caddy can also proxy to unix sockets !
does this work with multiple caddy servers? ie can you bind multiple caddy servers to port 80/443?
You can have multiple configs in a single Caddyfile and reload when you make changes, and it'll just route them as you wish, e.g.
domain1.com -> service on port 1234
domain2.com -> service on port 5678
domain3.com -> serving a file directory.
And then you still access domain1.com, domain2.com, domain3.com on port 80/443
You set up multiple services behind a single caddy reverse proxy
Btw, if you find yourself sending a lot of ports over ssh, you can also consider the option of having ssh start a socks5 proxy
ssh -D 4711 -q -C -N user@host
sets localhost:4711 up as a socks5 proxy you can tell your browser to use
...
A wireguard VPN is better of course; among other things because ssh is multiplexing over a single TCP connection and will encounter head of line blocking (where one dropped packet blocks all forwarded traffic until resent)
Love it!
I also did some experiments some time ago. The thing this is missing for me is the ability to also run arbitrary commands other that just using a few premade apps. In fact I think this stuff becomes really interesting when you put a real "shell" on top of this.
And I don't mean a classical posix shell, something that can be used to leverage the full power of the custom ui and frontend. Also a must have is "nestable connections".
The experiment I was doing was with a web interface and a statically compiled Go backend (for easy deployment via ssh). Maybe some day I will finish it xD
Do not do this. There are many, many excellent long-standing security and "web control plane isolation" reasons browsers are not permitted generic socket permissions.
The closest mechanical analog that comes to mind is why 3-wheeled ATVs are a bad idea.
I think it's okay as long as:
More here: https://outerloop.sh/security/Lovely writeup! I'll bookmark this for my own research.
My terminal's "clickity clackity" features [0] are local to the machine so I lose graphical-ness as soon as we remote in somewhere.
That's starting to change a bit with offline replay [1] where the native GUI and TUI work in tandem to unlock some rewind. But there's quite a road ahead and I love seeing others experiment properly. (Terminals are massively underserved.)
[0] https://terminal.click
[1] https://terminal.click/posts/2026/06/tui-stability/#:~:text=...
I think the approach here where interfacing with a device is considered from first principles is one that is rarely taken on, and this is a thought provoking implementation. Kudos.
This reminds me of an idea that I build a PoC of many years ago (maybe 2013 if I recall) that I always felt was the nugget of a useful idea. You would SSH into a server and processes on the other end would emit data which was then displayed in a webapp that was served from a localhost port, with a local backend that consumed the data. So for example a short-lived web-based remote 'top'. I did it as part of a company-internal hackathon and thought it was really cool, but nobody else was impressed with it. It was a very half-baked idea, and this looks like a fully-baked version of it. I'll check it out.
That's interesting idea. If we put into CLI with some ANSI escape code, that may become something real. Imagine a normal terminal app just render part of the UI in web and communicating in UNIX socket. While doing the fancy UI, everything is still controllable with keyboard, and optionally with mouse. The UI will fallback to text UI for older terminal
If your UI is not fully controllable with a keyboard, the same forces that made that happen will eventually make a mouse mandatory for this hypothetical tech stack too.
The terminal has no Platonic quality of being keyboard only. It is an accident of history and the limitations it has had. Remove the limitations and remove the accident of history and you will just end up drawn into the strange attractor of GUIs, warts and all.
There could be a brief honeymoon where the tech stack looks like some of you are imagining in your heads, but it would only last as long as it wasn't used by very many people. Google "gemini protocol" for a similar situation. That protocol has basically a cap on how popular it could possibly get before it just turned into HTTP B as the rest of the world forcibly upgraded it regardless of what the core project thinks. They exist in the shadow of HTTP, as the terminal exists in the shadow of GUIs. This is not a bad thing. It is what lets them be what they are. The shadows of GUIs or HTTP is large and there is plenty of space to be. Trying to give the terminal more GUI capabilities is like trying to give Gemini more web capabilities; you'll just end up in the same place, only with less refinement.
So, uh... X11? VNC? RDP?
No no not something on top of the UI stack. They also need framebuffer support so they are big headache to setup on headless server.
What I mean is that we can bring some web tech to terminal natively. We don't even need a separated shell. Security and bi-directional communication is built by default because of UNIX socket. But we still need to think how to handle stuff like cookie, local storage, external CSS / JS, ...
Web technologies are significantly larger and more complex than framebuffers, and they don't even let you start arbitrary programs like Chrome under them.
Interesting, kind of like a more fancy web shell. Haven't really ever seen the need for those, mostly because terminals work better than browsers.
Sometimes the browser is the only "computing platform" you have available (e.g. on some mobile devices, hotel kiosks).
Feedback: Home pages of each of Outer Loop, Outer Frame and Outer Shell contain basic intro of each instead of a link redirecting to them. By the time I click the link and on the new Outer X I have already what Outer X I came from and what it meant.
One of the more interesting pieces of Microsoft software is the Windows Admin Center where it's a web app to configure a Windows Server. Ideally, it was made for core installs where there's no GUI but it's there as a viable web management panel.
The tool from OP and WAC are pretty similar in terms of functionality and usecase. Why would you want this? Well, imagine your team needing to be able to do server functions but you have less technical team members to do it for you, which is very often the case in big places, most people are familiar with the web browser and having a website to do these sorts of actions makes it easier to have things done in one place without a lot of tools like Remote Desktop, SSH, WinRM, etc. configured.
At the risk of being considered a snob I don’t want someone who can’t deal with SSH or RDP configuring servers within my company. If you can’t work out how to SSH into the server you sure as hell aren’t going to work out how to safely expose network services on it.
Within your company, sure. But there's some engineers (think medical) who know standards like DICOM and PACS imaging but aren't familiar at all with OS internals or systems administration.
Lovely video and ingenious implementation. Kudos!
As someone managing various servers, both at home and at work, I see how this can be really useful. I see it not in the production space yet but rather in the experimenting, using a Linux machine as a second compute device!
So regarding your last point, I'm convinced. I think it is useful! The one fact that is bugging me is that now it requires a client specific app, with GUI, on my PC and I wonder if using ssh port forwarding could reduce the surface. I mean I wonder if either having a rich client that executes commands via ssh or a rich server (including Web Server) with ssh port wouldn't suffice, so that I can avoid installing stuff on the server AND on my computer.
Just had a quick look but I like the look so far. I’ve been thinking along similar lines for ages but never quite got around to making something. I very much support any effort to make remoting less dependent on the archaic character grid.
I’m good with just tailscale and self-hosted web-apps. Seems the main selling point is either native UX or reduced barriers to entry security-wise. I like barriers to entry.
This is cool. Though I don't see why someone would want to do more work/design for the custom GUI rendering for a custom/renderer (your viewer app) ?
I can’t make up my mind if I love it or hate it. On one hand this is like SSHapi on the other there’s no structure, no contract… i had similar doubts with Cockpit.
It's a cool video and I like the idea in general. The author mentions that the code runs in a sandbox. I'm surprised that WASM hasn't come up. You want the code to be platform agnostic anyway (it should run whether you start Outshell on Linux, macOS or whatever on different CPU architectures).
Thanks :)
I wrote a previous blog post that discussed WASM in the FAQ: https://probablymarcus.com/blocks/2026/05/10/like-a-web-view...
Being able to initiate a shell app from a regular remote ssh CLI prompt (like "ApacheConfig myhost.com" or "Editor ~/myrepo") might improve integration with people's existing CLI workflows.
It does need an agent that starts with every X or Wayland session and waits for requests from remote SSH sessions to start an app.
I don't really know what outerframe frame is. I tried to understand from the video and the blog but I'm still not sure what it is. Is it like a web browser but instead of DOM, HTML and JS you have Swift and SwiftUI running in a sandbox?
If so how would that work on non Apple devices? Also how much will that sandbox protect you?
It's purportedly cross-platform. The documentation leaves a lot to be desired, but it is described more here:
https://outerframe.org/ and https://outerloop.sh/native-apps/
Also a blog post about it, with its own video: https://probablymarcus.com/blocks/2026/05/10/like-a-web-view...
It's a fun heretical idea, moving away from a "cross-platform" web to a "multi-platform" web. It's a cross-platform protocol that hands off to platform-specific frontend code. I think it's a natural direction for the web, in a world where LLMs can translate to other platforms.
I thought this looks interesting, but was a little confused with what appears to be MacOS-only support at https://outerloop.sh/? I'm running Ubuntu 24.04, I kind of assumed from context that it'd be something I could spin up in a few minutes just to give it a go?
Also worth noting, my decision to give it a go relied mostly on the fact that I couldn't quite work out what the product is. Having "Outer Shell" and "Outer Loop" described as distinct-but-connected entities is a little confusing, IMO, which do I need to install, on what, and in what order?
Cool idea anyway, no shade here.
I have also been having trouble grasping the difference between Outer Loop and Outer Shell. I thought maybe one was the desktop browser app for macOS and the other was something running locally on the Pi to create the socket. However, after bouncing between the links for the two, I don't think that assumption was correct.
I am not sure I'd use this over exposing websites with wireguard as those will automatically work across platforms. But it looks like you could create some really cool experiences with it, and I'm happy people are exploring this space.
I wrote an early version of the Cylance AV desktop client. The UI side was a web app that talked to its windows service backend using HTTP over windows pipes. This was surprisingly easy to do using WCF.
I'm confused -- does this compile it live when the server ships code? How do we resolve dependencies, toolset etc.. Is the idea to just pick an old enough platform toolchain you expect to be present?
In all cases, the code is pre-compiled. A user never waits for anything to compile. When Outer Loop installs Outer Shell, it downloads pre-compiled binaries to the server. For Linux these are compiled against a manylinux ABI. Ditto for when Outer Shell installs one of the bundled apps. When a backend serves a native "web" app over HTTP it sends already-compiled ARM (or x86) code to the client.
Dependencies are less of a concern for the frontend binaries. For backends, I use a dependency-light approach, static-linking anything that's needed. Of course, people are welcome to do backends however they want, and just tell Outer Shell about the systemd/launchd units via the API. I used this no-dependency approach to keep everything lightweight and to keep install steps trivial, but admittedly it pushes me in certain directions (for example, using custom binary formats rather than sqlite).
I prefer hytelnet and MUDs but I don’t count, I’m just too old.
So a bit like X-forwarding used to do? Cool.
So.... webmin
Heresy!
I'm actually way more interested in option 2 - the VNC-like experience.
TUI apps are convenient over SSH because they're right there in your terminal. But they suck because they're restricted to shitty monospaced character grids. Why can't we have something more like VNC over SSH? Like, `top` and `micro` but with good graphics?
I did try doing something like that with the Kitty graphics protocol and you can get kind of close..ish, but it's really restricted by having to send everything as PNGs.
Anyway upvote for not being blinkered and thinking terminals are just for CLI stuff and must be forever.
In general I would like to see a web browser escape sequence for console applications. Just send a command to the terminal to connect a web browser to your stdin/out and present any UI you want over html. The terminal can then open a regular socket listening on localhost and act as a CGI server. For security the terminal should pick a random IP in the localhost range and a random URL. Technically that is security by obscurity, but guessing a cryptographically secure URL should be hard enough for attackers. The reasons to do it as an escape sequence and not just have the application open a socket and start the browser are: To enable remote GUI; To avoid the complexity of each application implementing networking; To enable better desktop integration, since the terminal itself is part of the Desktop Environment, so it can start a DE-specific browser, preferably in single-application mode. Also, it should be possible to automatically put the application in the background so you basically just run GUI applications like normal.
not sure what problem this solves. smells like setup for a security exploit. tho my most generous take is its CV-ware
> Isn’t it weird that this doesn’t already exist?
It does. MobaXterm have a bunch of it already, file manager on the side and ability to pass X11
> every app is a small HTTP server
This adds unnecessary overhead for communication. using web and web-like approaches on desktop system is a terrible idea.
Edit: withdrawing this objection, had no idea that right-clicking allowed the speed to be adjusted.
Sure, I just added YouTube mirror link to the post: https://youtu.be/e40PLLuZ5KI
(The one on the website is the standard browser video player, not custom.)
Thanks (and to pelzatessa as well), TIL about the right-click menu on these. That'll come in handy.
but its just standard <video> element, in firefox I can even right-click to change the speed to 2x. It's certainly better privacy-wise.
this exists as https://xpipe.io/
Cool, I hate it.
Defeats the purpose of the shell. The shell is for CLI interaction.
No. A shell is any user interface. Windows shell is explorer.exe and it used to be possible to change that via a config line in a system INI file.
SSH protocol also isn’t just for CLI work. It supports file transport (eg SFTP), TCP/IP forwarding and even SOCKS HTTP proxying.
You also used to be able to run GUI applications over SSH via X11.
You have a very loose definition of a shell that conflicts with about 40 years of history.
I don't have a dog in this fight, and anyway dogfighting is bad, but the intro to the Wikipedia article[0] reads:
> An operating system shell is a computer program that provides relatively broad and direct access to the system on which it runs. The term shell refers to how it is a relatively thin layer around an operating system.
> Most shells are command-line interface (CLI) programs. Some graphical user interfaces (GUI) also include shells.
The last line I think supports the notion that the term "shell" at least implies a CLI, but I can understand both positions.
---
0. https://en.wikipedia.org/wiki/Shell_(computing)
Edit: I'm shite at formatting on HN
The earliest versions of MacOS, all the way up through 9, had a ROM call at 0xA9F4 which was labeled `_exitToShell`. In the days before pre-emptive multitasking, this instruction's job was to force the current application to close and return the user to the MacOS desktop (the Finder). The "shell" in this context being the desktop user interface.
Just FYI.
I wondered if this would be controversial. It all depends where you grew up.
> Cairo, like Chicago, had a new shell (Microsoft’s favorite word for the user interface for launching programs and managing files) and a new file system
https://hardcoresoftware.learningbyshipping.com/p/020-innova...
When I worked at Microsoft 2010 - 2014, the word "shell" was still used in this way. I decided to say "graphical shell", to make it clearer.
Why do you think GNOME Shell is called GNOME Shell?
https://gitlab.gnome.org/GNOME/gnome-shell
(just as one example)
38 years ago: https://en.wikipedia.org/wiki/DOS_Shell
Not really no. I’ve been using shells and authoring new ones for around 40 years across a variety of platforms. The term has always been pretty loosely defined because as technology evolved the term “shell” was borrowed. So like I said, a shell can refer to a graphical core just as much as a text-based one. You can get web shells too.
The original intent was that a shell is a thin wrapper on top of the OS to expose the hosts capabilities. But that hasn’t been an apt description for most of those 40 years.
Appeal to authority.
command line shell vs graphical shell. My first experience with a graphical shell was dosshell[1]. For a while we called the Windows 3.1 interface "the shell". I guess the terminology has changed since that time.
[1] https://en.wikipedia.org/wiki/DOS_Shell
UI/UX is very bad why would we need it over Warp / Wave Terminal