I’m a Fedora KDE user and I missed two tools in a native Qt form, so I created them and would love feedback.
konqix is a Qt 6 instant messenger built on libpurple (no GTK), so it reuses libpurple’s protocol plugins such as XMPP, IRC and the Telegram tdlib plugin. It has tray integration, inline images and a searchable history, and it uses KF6::WindowSystem when available. It is licensed under GPL-2.0-or-later.
Konqix is an independent UI on the libpurple library and is not affiliated with the Pidgin project.
maria-sql-qlient is a lightweight Qt 6 query client for MariaDB and MySQL. It offers saved connections, a schema browser, multiple query tabs and history. It is licensed under GPL-2.0-or-later.
Maria-sql-qlient is not affiliated with MariaDB Corporation or Oracle.
Both build with CMake against Qt 6 on Fedora 43 (v0.1.0 tagged).
My question: what is the recommended path to take these toward beta testing within the KDE ecosystem? Is the KDE Incubator, or hosting on invent.kde.org, the right next step, and what are the prerequisites? Any feedback on the apps themselves is very welcome too.
I haven’t looked at the code but one commit each, co-authored by Claude, doesn’t inspire much confidence. The process of adding a new KDE project is documented here:
Let me be blunt. I have 40 years of programming experience and several commercial projects behind me. I started using AI as an assistant a few months ago, and I review every single line it writes. I do not ship code I do not understand, and I spent two weeks debugging and testing both apps myself. I credited Claude out of honesty, not because the code is a mystery to me.
The ‘spying’ and ‘stolen data’ claims are pure speculation from people who, by their own admission, have not read the code. Both apps are GPL and fully open source. konqix is a local libpurple frontend with no telemetry. It sends nothing anywhere except through the chat protocols you configure yourself, exactly like Pidgin. If you suspect otherwise, read the source and show me where. Until then, judging an app by the word “Claude” in its credits is prejudice, not a security review.
I’m happy to take concrete, technical feedback. Vague AI panic I’ll pass on.
A quick follow-up: listing Claude as a co-author was a mistake. Claude was not a co-author in any real sense, only an assisting tool. I have corrected how its use is described in the README and AUTHORS files: I am the author and maintainer, I have reviewed and understand every line, and I take full responsibility for the code.
I would really appreciate it if someone could try out these two products. I’m really not interested in silly images or irritable developers who are afraid that artificial intelligence will take their jobs.
That may be, but it’s not visible in the commit history. Of course all projects have to start somewhere, but I wonder if it’s a little premature to be talking about incubating these into KDE.
Thank you for your reply. I didn’t want to fill the commit history with early alpha testing. I now actively use konqix myself for company communication, and maria-sql-qlient whenever I need to verify something in the database. Over the past few days, everything I need has been working without failures. I can finally send attachments to my colleagues without having to use the web version of Telegram.
I assumed this forum might be a good place for the initial testing of applications before they go to invent-kde-org. So far it feels a bit like an AI-panic-attack support group here.
