🬤 rns.recipes

Forum / Showcase / Thread / Export

# Chat client focused on channel type communication

_Showcase · started by Hunter on Tue, Sep 1, 2026 10:55 PM_

---

## Original post

**Hunter** · Tue, Sep 1, 2026 10:55 PM

Im working on a side project that aims to bring discord style communication to reticulum. Im developing it using claude code as I have very little time to develop outside of my 9-5 as a SWE, family needs etc... That being said I very much am trying to avoid developing AI slop and am putting a lot of effort into creating an actually functional app and not just half assed AI slop. Im hoping to get some honest feedback and also check in to see if this is something being attempted elsewhere. Happy to provide a link to the github project, discuss how im designing the system etc... First time posting here as well so hopefully I haven't broken a rule by posting this

---

## Reply 1

**Boltic** · Tue, Sep 1, 2026 11:11 PM

This is very good, maybe you can use free software for it with a little modification? For example, mumble, of course, it's old in appearance and because it's based on IP, it's probably a bit difficult, but it's a complementary thing (:

Anyway, I always thought the same and I was more concerned with the convenience of the IP transmission tunnel than creating a separate program, which is a good thing, but the bad thing is that you can only make a program for the reticulum network and everything has to be developed for it... Of course, it can support both IP and reticulum, as far as I'm concerned now it doesn't matter (:

Anyway, we're waiting! And I hope it doesn't become something that people are afraid to use and say, wow, AI has woven something for itself and what else (: Good luck, I'm waiting for this thing...

---

## Reply 2

**Hunter** · Tue, Sep 1, 2026 11:32 PM

voice is something ive started to mess with but im hesitant due to the bandwith implications of multiple people all talking in a shared voice channel. My main focus is nailing text based communication in a shared channel for now.

---

## Reply 3

**Boltic** · Tue, Sep 1, 2026 11:38 PM

**Hunter** wrote:

> voice is something ive started to mess with but im hesitant due to the bandwith implications of multiple people all talking in a shared voice channel. My main focus is nailing text based communication in a shared channel for now.

Yes, that's the case, but with the 8kb opus codec, which is one kilobyte, if several people are talking, and we consider that one person in the middle plays the role of transmitting the sound to everyone, it doesn't consume much bandwidth.

Anyway, that's good for a start, and the rest is your goal (:

---

## Reply 4

**Zenith** · Tue, Sep 1, 2026 11:44 PM

There is LXST which is supported by Sideband, Columba, and MCX for voice. It uses both Opus and codec2 for audio codecs

https://github.com/markqvist/lxst

And RRC for group chats which has support in Nomadnet, Columba, and MCX

https://github.com/kc1awv/rrcd

Group chats with multimedia support just isn't possible to do over Reticulum when considering slower interfaces and network topologies. RRC is the simplest to get up and running and leaves a lot of MTU overhead for the message content itself and already has existing support in many LXMF clients

I would say contribute to existing clients and protocols we already have that are in use today, supported by developers, instead of vibecoding another one.

---

## Reply 5

**Anonymous** · Tue, Sep 1, 2026 11:44 PM

Curious on the architecture, is this intended to be deployed similarly to ventrilo or teamspeak in the days of old?

---

## Reply 6

**Hunter** · Wed, Sep 2, 2026 12:07 AM

from what i understand rrc is meant to function like irc. My goal is something of a decentralized but private channel where members who are offline can reconnect and update their local version of the channel by syncing with other members who have the missing content. RRC looks like it requires the host node to be online, otherwise the channel goes down

---

## Reply 7

**Zenith** · Wed, Sep 2, 2026 12:16 AM

**Anonymous** wrote:

> Curious on the architecture, is this intended to be deployed similarly to ventrilo or teamspeak in the days of old?

You could do this today with LXST. There's a built in Mixer primitive that can combine input sources for multiple speakers and you would just need to send it down over a bridge. And build some type of simple client that has push to talk or voice activity detection.

---