[Feature Request] expose port for MC app to use #131

New Issue

2026-10-05T01:59:01+02:00

adam commented

2026-10-05 01:59:01 +02:00

Hi
Great work on the app!
Just started using it and have few requests to make =)
Wanted to start with asking if it's possible to expose an additional port in the app, so that Meshcore client app can connect to and be used like normal?
It's great that web app chat is available but its not sending alerts when new message arrives etc.
Plus used a MC app layout =)
The plan was to use my esp32 node as a remote companion, if I do not have my physical companion device with me. But with remote term connected, means that I cannot use the app anymore, only Web app
Should not be too hard (theoretically)to make it talk to the companion if someone connected to the remote term via the app on say port 5001 or something. So it will also become a bridge for devices that are usb/ble/wifi connected through the remote term address

Hi Great work on the app! Just started using it and have few requests to make =) Wanted to start with asking if it's possible to expose an additional port in the app, so that Meshcore client app can connect to and be used like normal? It's great that web app chat is available but its not sending alerts when new message arrives etc. Plus used a MC app layout =) The plan was to use my esp32 node as a remote companion, if I do not have my physical companion device with me. But with remote term connected, means that I cannot use the app anymore, only Web app Should not be too hard (theoretically)to make it talk to the companion if someone connected to the remote term via the app on say port 5001 or something. So it will also become a bridge for devices that are usb/ble/wifi connected through the remote term address

adam closed this issue

2026-10-05 01:59:01 +02:00

adam commented

2026-10-05 01:59:02 +02:00

Thank you!

Unfortunately, that's not something that's possible — a MeshCore radio expects one client to be connected to it, speaking the wire protocol for radio control. RemoteTerm is subscribed to radio events, mutating radio state, loading and offloading contacts, etc., all actions that a client is going to assume is NOT happening underneath them (they would expect that the radio is in the last state it was used in). It might be possible to set up a passthrough where we drop all our subscribed radio actions and just pass through traffic, but that totally defeats the purpose of remoteterm, which is that it runs all the time and ingests 100% of mesh traffic. If that was desired, then it would make more sense to use the in-app Disconnect button and just use the radio directly. Proxying a byte-level protocol is far more reliably done by dedicated proxy apps like socat rather than hacking remoteterm to do it.

Thank you! Unfortunately, that's not something that's possible — a MeshCore radio expects one client to be connected to it, speaking the wire protocol for radio control. RemoteTerm is subscribed to radio events, mutating radio state, loading and offloading contacts, etc., all actions that a client is going to assume is NOT happening underneath them (they would expect that the radio is in the last state it was used in). It *might* be possible to set up a passthrough where we drop all our subscribed radio actions and just pass through traffic, but that totally defeats the purpose of remoteterm, which is that it runs all the time and ingests 100% of mesh traffic. If that was desired, then it would make more sense to use the in-app `Disconnect` button and just use the radio directly. Proxying a byte-level protocol is far more reliably done by dedicated proxy apps like socat rather than hacking remoteterm to do it.

adam commented

2026-10-05 01:59:03 +02:00

Yeah, I briefly ran my ideas through Claude and it gave some ideas, but not sure if they can be implemented/will work
My idea was that mobile MC app connects to remote term on another port, thinking its a companion, and some daemon translated app requests into the requests that remote term already uses- like get messages, send messages etc

My main gripe is with notifications on mobile, which do not work. Can this be made to run in its own app (for some sites I see installable Chrome apps) or something like remote terminal chat, with reliable notifications?

Yeah, I briefly ran my ideas through Claude and it gave some ideas, but not sure if they can be implemented/will work My idea was that mobile MC app connects to remote term on another port, thinking its a companion, and some daemon translated app requests into the requests that remote term already uses- like get messages, send messages etc My main gripe is with notifications on mobile, which do not work. Can this be made to run in its own app (for some sites I see installable Chrome apps) or something like remote terminal chat, with reliable notifications?

adam commented

2026-10-05 01:59:03 +02:00

Progressive Web app

Progressive Web app

adam commented

2026-10-05 01:59:04 +02:00

Yeah I've had thoughts about using Web Push (we are already PWA compatible); I have an experimentation branch I should push out and get some beta feedback on. I'll keep you posted.

Yeah I've had thoughts about using Web Push (we are already PWA compatible); I have an experimentation branch I should push out and get some beta feedback on. I'll keep you posted.

adam commented

2026-10-05 01:59:05 +02:00

Great!
How do I use PWA? Get no option

Great! How do I use PWA? Get no option

adam commented

2026-10-05 01:59:05 +02:00

If you're already serving over a trusted context, it should just be as simple as using the install prompt from your browser menu. However, you need to have a trusted HTTPS context which often needs per-browser customization to add your RemoteTerm host to an always-trusted list (or you can use a non-snakeoil HTTPS cert).

If you're already serving over a trusted context, it should just be as simple as using the install prompt from your browser menu. However, you need to have a *trusted* HTTPS context which often needs per-browser customization to add your RemoteTerm host to an always-trusted list (or you can use a non-snakeoil HTTPS cert).

adam commented

2026-10-05 01:59:06 +02:00

Ah sorry looks like I never actually pushed up the serviceworker required for that to work. Doing now.

Ah sorry looks like I never actually pushed up the serviceworker required for that to work. Doing now.

adam commented

2026-10-05 01:59:06 +02:00

Okay that's pushed. Now, if you have a trusted context (i.e. valid non-self-signed cert), PWA will work.

Okay that's pushed. Now, if you have a trusted context (i.e. valid non-self-signed cert), PWA will work.

adam commented

2026-10-05 01:59:07 +02:00

I'm gonna close this for now, but if you run into any additional issues, feel free to open a new one!

I'm gonna close this for now, but if you run into any additional issues, feel free to open a new one!

adam commented

2026-10-05 01:59:08 +02:00

Using docker. Would it automatically update?

Using docker. Would it automatically update?

adam commented

2026-10-05 01:59:08 +02:00

How do you mean? You'll need to configure a TLS certificate using a tool like Certbot or a secure tunnel like cloudflared. You can pass the certificate into Docker or use a tunnel that passes traffic to Docker. Unfortunately, setting up trusted HTTPS can be a bit technical if you haven't done it before; you might be able to find some tutorials to help.

I think you can pass a command line flag to Chrome to treat an insecure origin as secure but that will only work for desktop browsers.

How do you mean? You'll need to configure a TLS certificate using a tool like Certbot or a secure tunnel like cloudflared. You can pass the certificate into Docker or use a tunnel that passes traffic to Docker. Unfortunately, setting up trusted HTTPS can be a bit technical if you haven't done it before; you might be able to find some tutorials to help. I think you can pass a command line flag to Chrome to treat an insecure origin as secure but that will only work for desktop browsers.

adam commented

2026-10-05 01:59:09 +02:00

You said that you just pushed the code, I assume github. How does it update docker image at docker.io?

You said that you just pushed the code, I assume github. How does it update docker image at docker.io?

adam commented

2026-10-05 01:59:10 +02:00

Ohhhhh I follow. Yeah I'll push up a build in 20 minutes or so.

Ohhhhh I follow. Yeah I'll push up a build in 20 minutes or so.

adam commented

2026-10-05 01:59:11 +02:00

Thank you =) somewhat new to the dev work

Thank you =) somewhat new to the dev work

adam commented

2026-10-05 01:59:11 +02:00

No worries; sorry for misunderstanding your question! New build is pushed.

No worries; sorry for misunderstanding your question! New build is pushed. ![Screenshot_20260412-205600.png](https://github.com/user-attachments/assets/56349876-4a51-40ec-9f12-c3d1b6352d6e)

adam commented

2026-10-05 01:59:12 +02:00

Much appreciated =)
If I may use your knowledge a bit more - I am using tailscale for remote access and it generates the certs.
When I just generated them and updated docker compose file and restarted it, it just would not want to use the certs until I deleted the container and rebuilt

Much appreciated =) If I may use your knowledge a bit more - I am using tailscale for remote access and it generates the certs. When I just generated them and updated docker compose file and restarted it, it just would not want to use the certs until I deleted the container and rebuilt

adam commented

2026-10-05 01:59:13 +02:00

Hmmmm interesting. If it's tailscale handling the certs then I wouldn't expect there to be a need to delete and recreate, unless there's something linked to the container like a tailscale container that's handling the proxying. Even then I wouldn't expect a destroy to be necessary. On the upside, the data directory should be mounted which means that destroying and recreating doesn't really make a material difference to anything/all the data should still be retained.

Hmmmm interesting. If it's tailscale handling the certs then I wouldn't expect there to be a need to delete and recreate, unless there's something linked to the container like a tailscale container that's handling the proxying. Even then I wouldn't expect a destroy to be necessary. On the upside, the data directory should be mounted which means that destroying and recreating doesn't really make a material difference to anything/all the data should still be retained.

adam commented

2026-10-05 01:59:13 +02:00

Yeah, tailscale is just handing cert file updates. But when app was reloaded it did not use them
Anyway, thank you for your help

Yeah, tailscale is just handing cert file updates. But when app was reloaded it did not use them Anyway, thank you for your help

adam commented

2026-10-05 01:59:14 +02:00

Huh, weird. Not sure what would stop it from picking up new files unless they've moved or something.

Huh, weird. Not sure what would stop it from picking up new files unless they've moved or something.

adam commented

2026-10-05 01:59:14 +02:00

Ok. Reloaded config. Install still does not work. When inspecting the code, I get 401 error for the webmanifest file

Ok. Reloaded config. Install still does not work. When inspecting the code, I get 401 error for the webmanifest file

adam commented

2026-10-05 01:59:15 +02:00

hm have you set up basic auth? That would be the only way I'd expect the app to return a 401. Maybe wrong username/password or your browser needs to reauth? Are any other resources 401ing or just the webmanifest

hm have you set up basic auth? That would be the only way I'd expect the app to return a 401. Maybe wrong username/password or your browser needs to reauth? Are any other resources 401ing or just the webmanifest

adam commented

2026-10-05 01:59:16 +02:00

Yes, basic authentication, as recommended
Other components load up fine apart from manifest
Ai suggest updating index file and some backend code that handles it

Yes, basic authentication, as recommended Other components load up fine apart from manifest Ai suggest updating index file and some backend code that handles it

adam commented

2026-10-05 01:59:17 +02:00

Aha, I found the bug. Will push up a fix.

Aha, I found the bug. Will push up a fix.

adam commented

2026-10-05 01:59:17 +02:00

Okay new docker image with that bugfix is uploaded. With that, I am on vacation! I will be back at work on remoteterm on thursday or so.

Okay new docker image with that bugfix is uploaded. With that, I am on vacation! I will be back at work on remoteterm on thursday or so.

adam commented

2026-10-05 01:59:18 +02:00

I am surprised that you are so actively responding!
Thank you and enjoy!

I am surprised that you are so actively responding! Thank you and enjoy!

adam commented

2026-10-05 01:59:18 +02:00

(and sorry, to be clear, that was meant as "and now my vacation begins (yay)" not "I'm on vacation; stop bothering me") Always happy to answer questions as best I can, esp if they're bugs I wrote 😉

(and sorry, to be clear, that was meant as "and now my vacation begins (yay)" not "I'm on vacation; stop bothering me") Always happy to answer questions as best I can, esp if they're bugs I wrote 😉

adam commented

2026-10-05 01:59:19 +02:00

Thank you. 401 is fixed but still does not want to install.
Will wait until you are back at it later in the week

Thank you. 401 is fixed but still does not want to install. Will wait until you are back at it later in the week

adam commented

2026-10-05 01:59:20 +02:00

For sure! If you're using Chrome, right click -> inspect element -> Application (tabs at the top) -> Service Worker (left hand bar) should show you the status of service worker installation, which has been the limiting factor for installation as a pwa until now. That might have some interesting debug leads if you're feeling adventurous. Problems with installation might also be visible on the error console. But yeah, if you haven't figured it out, we could dive deeply into it once I'm back.

For sure! If you're using Chrome, right click -> inspect element -> Application (tabs at the top) -> Service Worker (left hand bar) should show you the status of service worker installation, which has been the limiting factor for installation as a pwa until now. That might have some interesting debug leads if you're feeling adventurous. Problems with installation might also be visible on the error console. But yeah, if you haven't figured it out, we could dive deeply into it once I'm back.

adam commented

2026-10-05 01:59:20 +02:00

I also tried accessing the manifest file without authentication, which is what browser expects I think. And it's unauthorised
When logged on, it loads fine, but it could be a browser requirements

I also tried accessing the manifest file without authentication, which is what browser expects I think. And it's unauthorised When logged on, it loads fine, but it could be a browser requirements

adam commented

2026-10-05 01:59:21 +02:00

The bug fix that I pushed up an image for should include the tag that tells the browser that it will need to pass the credentials on in order to access that file; if you are still seeing a 401 in your network requests, clearly something's going screwy there.

The bug fix that I pushed up an image for should include the tag that tells the browser that it will need to pass the credentials on in order to access that file; if you are still seeing a 401 in your network requests, clearly something's going screwy there.

adam commented

2026-10-05 01:59:22 +02:00

Ai says manifest must be accessible without auth. I will try removing basic authentication and test

Ai says manifest must be accessible without auth. I will try removing basic authentication and test

adam commented

2026-10-05 01:59:22 +02:00

adam commented

2026-10-05 01:59:23 +02:00

Now AI says that even though we have

"icons": [
{
"src": "https://smartserver.tail7e6204.ts.net:8000/web-app-manifest-192x192.png",
"sizes": "192x192",
"type": "image/png",
"purpose": "maskable"
},
{
"src": "https://smartserver.tail7e6204.ts.net:8000/web-app-manifest-512x512.png",
"sizes": "512x512",
"type": "image/png",
"purpose": "maskable"
}

at least of these is meant to have purpose of "any"

Now AI says that even though we have "icons": [ { "src": "https://smartserver.tail7e6204.ts.net:8000/web-app-manifest-192x192.png", "sizes": "192x192", "type": "image/png", **"purpose": "maskable"** }, { "src": "https://smartserver.tail7e6204.ts.net:8000/web-app-manifest-512x512.png", "sizes": "512x512", "type": "image/png", **"purpose": "maskable"** } at least of these is meant to have purpose of "any"

adam commented

2026-10-05 01:59:24 +02:00

Added and pushed 🙂

Added and pushed 🙂

adam commented

2026-10-05 01:59:24 +02:00

It did not like svg icons. But even when manually removed, it's still not giving me option to install. Giving up for now

It did not like svg icons. But even when manually removed, it's still not giving me option to install. Giving up for now

adam commented

2026-10-05 01:59:25 +02:00

After a while (and don't know if a coincident) it prompted to install! (I removed svg icon)

After a while (and don't know if a coincident) it prompted to install! (I removed svg icon)

adam commented

2026-10-05 01:59:26 +02:00

Awesome; I've moved those over to pngs. And yeah, I think the prompt is based on a wall clock of usage time but it should always be addable from the browser menu. I'm gonna close this for now; if you run into any other PWA issues feel free to open a new issue.

Awesome; I've moved those over to pngs. And yeah, I think the prompt is based on a wall clock of usage time but it should always be addable from the browser menu. I'm gonna close this for now; if you run into any other PWA issues feel free to open a new issue.

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: adam/Remote-Terminal-for-MeshCore#131