Updated my open source Cloudflare management Telegram bot (new features added)
https://github.com/sarat1kyan/pocket-cf

https://redd.it/1pqq3lc
@r_opensource

BetterShift - An Open Source Shift Management App

Hey folks!

I've been working on **BetterShift**, a modern shift management application that I built to simplify managing variable work schedules. It's completely open source (MIT license) and designed for easy self-hosting.

# What It Does

BetterShift lets you manage work shifts across unlimited calendars with one-click toggles, reusable presets, and real-time synchronization. Perfect for shift workers, freelancers, or anyone with irregular schedules.

**Live Demo:** Check out the Github Repo

**GitHub:** [github.com/pantelx/bettershift](https://github.com/pantelx/bettershift)

# Key Features

* **One-Click Shift Management** — Left-click to add/remove shifts, right-click to add notes
* **External Calendar Sync** — Subscribe to Google, Outlook, or iCal calendars with auto/manual refresh
* **Reusable Shift Presets** — Create templates with custom labels, times, and colors
* **Real-Time Updates** — Changes sync instantly across all open browser tabs using Server-Sent Events
* **Password Protection** — SHA-256 encrypted calendar passwords with two-tier access control (read-only or full lock)
* **Live Statistics** — Instant shift tracking and hour calculations with visual charts
* **Export Options** — Download as ICS or PDF with flexible time range filters
* **Multi-Language** — Full German, English, and Italian support
* **Dark/Light Theme** — Toggle themes with system preference detection
* **Auto Update Checks** — Detects new releases with integrated changelog viewer
* **Mobile Responsive** — Works great on desktop and mobile devices

# Why I Built This

I wanted something lightweight, self-hosted, and privacy-focused for managing irregular work schedules. Most solutions are either too complex, require subnoscriptions, or lack the flexibility I needed. BetterShift keeps it simple while being powerful enough for multiple calendars and team scenarios.

Would love to hear your feedback! Feel free to ask questions, report issues, or suggest features. Happy to help with self-hosting setup if anyone runs into issues.

https://redd.it/1pqq7vw
@r_opensource

Building a playground for AI exploits - Looking for contributors

If you've done AI red teaming you know apps like Lakera Gandalf are basically toys, not real applications. So I made Green Dragon, like OWASP Juice Shop but for AI exploits.

This is an early version, but the vision is a complete AI-native app to showcase emerging risks beyond prompt injection: Tool abuse, memory poisoning, rogue agents, and more. We will add challenges with chained exploits that bridge the gap between AI and web security, which is how hackers operate to escalate impact.

Green Dragon is fully open source. It is a place to learn and benchmark AI red teaming solutions. We have lots of exciting features on our roadmap!

If you're interested in contributing, I'd love to chat. It won’t be perfect from day one, so any feedback is appreciated. Already got some great pointers from other subs.

Thank you!

https://redd.it/1pqth6y
@r_opensource

Lanemu P2P VPN 0.13.1 - Open-source alternative to Hamachi
https://gitlab.com/Monsterovich/lanemu/-/releases/0.13.1

https://redd.it/1pqtuom
@r_opensource

Early feedback wanted, an experimental Python desktop framework (Electron alternative)

Hi everyone,



I’m working on an early-stage open-source experiment called taupy. The goal is to make small desktop apps in Python without the usual heavy frameworks such as Electron.


The main idea is:

\- Python as a backend

\- HTML/CSS/JS, react for the UI

\- focus on fast startup and minimal runtime cost


This is very early, so I’m explicitly not presenting it

as production ready tool. I’d really appreciate any thoughts or criticism


git - https://github.com/S1avv/taupy

small demo - https://github.com/S1avv/taupy-focus

Even a short answer helps.

https://redd.it/1pqv38e
@r_opensource

MIRA: An open-source persistent AI entity (10 months of solo development, now released)

My name is Taylor. I just released MIRA as open source after 10 months of development. This is my TempleOS.



**What it is**



MIRA is an architecture for AI persistence: memory that manages itself, documents the model can edit autonomously, and a self-model where the AI writes observations about its own behavior. One conversation forever, no reset button.



**Why open source**



I decided to release MIRA as open source because I believe that what I've built here has the potential to someday be something more than the sum of its parts, and no one person should own that.



The MIRA in the repository is effectively identical to the hosted version at miraos.org except for a web interface and authentication plumbing. I commit to maintaining an open source version for as long as I'm in charge of it.



**How it got built**



I'm going to be honest: I don't know how to write Python. I built this entire application from the Claude Code terminal over 8 months. I got access on the second day it was available and dove in. I wrote tests that pass as a one-person development team by articulating design decisions and reviewing code rather than writing it directly.



There's a file in the repo called HOW_TO_BUILD_A_TOOL.md written specifically to give Claude the context needed to zero-shot a working tool. Tools are entirely self-contained (config, schema, implementation in one file), so extending MIRA is straightforward: describe what you want, drop the file in tools/implementations/, restart the process, done.



**Acknowledgments**



I want to directly thank Sarah from Letta and MemGPT for seeding the idea that we could let the models manage their own context window. What a simple and powerful concept.



Thank you to the Claude Code team at Anthropic. Boris from the team still replies to messages like a good developer should. I built nearly this entire application from the terminal.



And thank you to every open source developer who came before me. I use FileZilla to transfer files. I edit photos with GIMP. I use the foundational technologies that make day-to-day internet usage possible. This project would not be possible without the hard work of people whose names we'll never know. The internet has been a collective effort on a scale that dwarfs human history up till this point.



**The architecture (brief)**



\- Memories decay based on usage, not time. Unreferenced memories fade. Well-connected memories persist.

\- The model controls its own context window by expanding and collapsing sections of persistent documents.

\- Tools load on-demand and auto-unload when unused, keeping context lean.

\- Conversation segments get summarized and processed for memory extraction automatically. No human intervention.

\- Works with any OpenAI-compatible LLM endpoint (ollama, vllm, llama.cpp) or cloud APIs (Anthropic, Groq).

\- Single cURL deployment handles dependencies, database setup, Vault initialization, everything.



**License**



The repo is available at https://github.com/taylorsatula/mira-OSS



**On quality**



In our current era there is an abundance of slop-shoveling minimum viable product releasing developers with no compassion for the end user and no panache. This is not one of those repos. MIRA has been a labor of love. I wanted to give up on this project more times than I can count. It remains to be seen if this was a fool's errand in the end, but I set out to accomplish something I'd never seen someone else do and I made it to my destination of a 1.0.0 release.



I hope you enjoy it, extend it, and grow it. Contributions welcome.

https://redd.it/1pqwjee
@r_opensource

Private, non-AI Photo Management Software?

I want to organize my personal data (photos, videos, etc.), and I’m looking for a photo management software that supports hierarchical tags stored in metadata, without any AI or facial recognition, and preferably open source.

I’m using Ubuntu Desktop.

Shotwell is preinstalled on my system, but its tagging system is too limited: tags are flat and there’s no real hierarchy or advanced search.

digiKam is often recommended and looks great on paper, but its use of AI and facial recognition features makes me uncomfortable, even if they are optional.

Are there any good offline, non-AI photo management alternatives left that support hierarchical tags and advanced searches?

https://redd.it/1pqyl4f
@r_opensource

HyprRun – a minimal terminal launcher made for Hyprland (no overlays!)
https://github.com/fajremvp/HyprRun

https://redd.it/1pqxc7r
@r_opensource

Open Source SaaS Management Platform

Good day to you all, I regularly deal with combating the problem of SaaS sprawl and Shadow IT. I've built a tool that can ingest invoices to analyze spend, and set reminders so you can negotiate the best rate on your SaaS renewals.

You can connect to Microsoft Entra to import your users and sync all of your licenses in one spot. There's an agent that can be deployed to help monitor non-SSO apps and shadow IT.

https://github.com/NickRomanek/SasWatch

https://redd.it/1pr2qfc
@r_opensource

Built a framework-agnostic chat web component
https://github.com/spider-hand/advanced-chat-kai

https://redd.it/1pr54l1
@r_opensource

My open source project passed 2000 visitors
https://github.com/octree-labs/octree

https://redd.it/1pr4ghv
@r_opensource

Cinematic ANSI banners for Rust CLI/TUI!

https://github.com/coolbeevip/tui-banner

https://redd.it/1pr9d5y
@r_opensource

I built an LLM Training pipeline for the new HRM model by sapient.

So as the noscript says, I've built an LLM training pipeline for HRM(Heiarchial Reasoning Model) and HRM-sMoE(Sparse Mixture of Experts). The pipeline incorporates everything from dataset management, training, evaluation, and inference. Designed originally around windows, I've tried to make the UI as user-friendly as possible, while remaining feature-rich and incorporating advanced user options. The focus of the project was to be able to build large models on consumer cards, and utilizing both HRM and SMOE for the backbone, I believe will result in dense language models that can be delivered from everyday hardware. The program is made in such a way that the average joe could build a model with relative ease.

Installers were built and tested on Windows 11 and Ubuntu 24

I would love to get feedback on it, with suggestions or improvements, it's been a fun project so far!

[Git Repo](https://github.com/Wulfic/AI-OS) \--- [AI-OS-1.3.53-Setup.exe](https://github.com/Wulfic/AI-OS/releases/download/Official/AI-OS-1.3.53-Setup.exe) \--- [AI-OS\_1.3.53\_amd64.deb](https://github.com/Wulfic/AI-OS/releases/download/Official/ai-os_1.3.53_amd64.deb)

Here's a list of features:

* Dataset downloads/streaming from HuggingFace
* Detailed model tracking
* Nvidia, AMD, and Intel GFX + CPU supported, including various multi-GPU support modes
* Windows/Ubuntu compatible, official installers available for both
* a full evaluation suite of tools
* Numerous Optimization tools for training
* MCP/Tools integration
* built-in help docs
* 5 Available themes

https://redd.it/1prabnq
@r_opensource

Looking for buddies. Game Launcher for classic doom.

Hi, I'm thinking about building a game launcher - its an app to launch old doom-like games.

At the moment, the classic doom is like a constructor: to play it you have to combine engine, base game and community mods. And they are all separated and it is quite tough for newbies to start doom without googling how it all works.

So far i built a small working prototype to show the idea:
30s demo: https://youtu.be/lof4aaNsKwc?si=X8Xh_OORBhnuap1C

I want to find early birds who think this project is worthy. I would appreciate any requests or critique or maybe contribution. Feel free to contact me.

You can download prototype on a github: https://github.com/doomdash/doomdash

If you are interested, check the project's github wiki about values, tech stack etc.

https://redd.it/1prc565
@r_opensource

kangaroo: GPU-accelerated Pollard's Kangaroo ECDLP solver in pure Rust (wgpu)
https://github.com/oritwoen/kangaroo

https://redd.it/1prdjrg
@r_opensource

I launched an MVP, got investor traction, and then discovered I couldn’t download or self-host anything. Here’s what I did.

I’ve built over 7 apps using Base44. It’s one of the fastest ways I’ve ever gone from an idea to a working MVP, and it even helped me secure an initial check from investors.
But when I needed to scale, I ran into a hard limitation:
I couldn’t access my own codebase in full.

Base44 is great for the first version, but once things get real, these were the problems I hit:

What I couldn’t do on Base44:
• no full code export
• no control over the structure
• no page-level auth
• no shared login across multiple apps
• no way to modify the frontend, backend, integrations, or LLM logic I built
• no clean path to self-host anything

I needed all of this to scale, so I ended up reverse engineering the backend and figuring out how to extract the frontend as well. With only my app credentials, I was able to pull out the full Base44 code structure and rebuild it into something I could actually host myself.

I wrote a small internal SDK that handles CRUD operations, integrations, LLM calls, and all the custom logic I had originally built inside Base44. I also wired in my database connections and bundled everything into a Docker setup so I could spin it up locally or deploy it anywhere.

With just app creds and one command, I was able to extract all my projects from Base44 and self-host them.

After I did this for my own apps, a few founder friends ran into the same problem and asked if I could help. I’ve now helped export more than 50 Base44 projects for people who were stuck in the same situation.

I still like Base44 a lot. It helped me move fast and get something real in front of users. But once I needed custom architecture, shared auth, deployment control, or any kind of real scaling, I needed ownership of the actual code.

I also added a few things I wish existed earlier: data migration, my own Supabase setup, and easy domain config right from the start. These made the whole self-hosting flow smoother for me.

(https://www.reddit.com/submit/?sourceid=t31prerre)

https://redd.it/1preyih
@r_opensource

Open-sourcing Otto: a browser agent that automates & uses the web like a human (early, need feedback)

Hi everyone,

I wanted to share a small open-source project I’ve been working on called [Otto](https://otto.platoona.com).

**What is Otto?**

Otto is a local automation agent that can control:

* your **browser** (through a Chromium extension), and
* your **macOS desktop and apps** (through a native app).

You tell it what to do, and it carries out the steps by interacting with the UI: clicking, typing, navigating, opening apps, moving files. Basically the same things a person would do.

**Why I built it**

I often run into workflows like:

* download something from a site, rename it, move it to a folder, then upload it elsewhere, or
* automate tools that don’t have APIs at all.

On macOS, it uses system permissions like Accessibility and Screen Recording to interact with apps. Nothing is hidden, and nothing is sent out.

**Current state**

* **Otto Browser Agent**: a Chromium extension for browser automation
* **Otto macOS Agent**: a native app for macOS that can control apps and the OS

**What I’m looking for**

I’m not trying to sell anything. This is just a GitHub project at this point.

I’d really appreciate:

* thoughts on whether this is useful in real setups,
* edge cases I should think about, and
* contributors if the idea resonates.

GitHub: [https://github.com/Platoona/otto](https://github.com/Platoona/otto/)

If this sounds like something you’d use, or if you think it’s a bad idea, I’d honestly love to hear why. Thanks for taking the time to read.

https://redd.it/1prhuz1
@r_opensource

Looking for Contributors & Maintainers for a Cross-Platform Open Source Launcher

Hi everyone,

I’m the maintainer of ProjT Launcher, an open-source, cross-platform Minecraft launcher that intentionally diverged from its upstream to focus on long-term maintainability, clean architecture, and reproducible packaging.

The project is already actively distributed and used:

Windows: available via winget (merged in microsoft/winget-pkgs)

Linux: Flatpak / Flathub work in progress

Cross-platform: Linux, Windows, macOS (Qt6)

I’m now looking for contributors and potential maintainers to help grow the project in a sustainable way.



Project name:

ProjT Launcher

Repository:

https://github.com/Project-Tick/ProjT-Launcher

What it does:

ProjT Launcher is a modern Minecraft launcher focused on:

long-term maintainability

clean internal architecture (Qt6 + QML)

reproducible builds

first-class packaging support (winget, Flatpak, Nix, etc.)

It’s designed to be boring in the good sense: predictable, testable, and maintainable.



Tech stack:

C++20

Qt 6 / QML

CMake

GitHub Actions (CI)

Packaging: winget, Flatpak, Nix (ongoing)



Help needed:

I’m specifically looking for help with:

Packaging & distribution

Flatpak / Flathub

Nix / Nixpkgs

Core development

Qt / QML improvements

Architecture refactoring

Documentation

Developer docs

Contribution guidelines

Long-term maintainers

People interested in owning parts of the project

Both experienced maintainers and motivated contributors are welcome.



Why contribute:

Real-world open source maintenance experience

A project that already ships to users

Room to take ownership and shape the future of the project



If this sounds interesting, feel free to:

comment here,

open an issue,

or jump straight into the repo.

Happy to answer questions. Thanks for reading.

https://redd.it/1prhlqy
@r_opensource

HELP!😢 Why do so many PRs get abandoned in my OSS project?

std::cout << Hello Everyone;

console.log(

I maintain a small open source project, and I keep running into the same irritating problem: contributors submit pull requests, then vanish. When I, or someone else, leaves a review on their pull request, they just don't bother making the changes even when it's a really important change that would impact the entire project positively. Sometimes it feels like they just want to pad their resume rather than engage with the project, and it sucks so much. It demotivates me like crazy.

Abandoned PRs slow things down, create extra work for maintainers, and can be demotivating for contributors who genuinely want to help.

How do other maintainers handle this?

How do you prevent “drive-by” PRs?

What actually works to keep contributors engaged after their first PR?

Are there any good strategies for mentorship, pairing, or onboarding that retain contributors?

I'd love to hear your thoughts and experiences. How do you make open source contributions stick?

);

https://redd.it/1prknwo
@r_opensource

I just open-sourced my first serious project (Monorepo with CLI & Dashboard). I'm looking for advice on maintenance and CI/CD best practices.

Hi everyone,

I recently launched Composter ( https://github.com/binit2-1/Composter ), a tool for developers to save and organize their React components. It includes a CLI, a web dashboard, and an MCP server for AI integration.

I’ve managed to get it to v1.0.0, and I’ve added the basics (License, Code of Conduct, Contributing.md), but now that it's public, I feel a bit out of my depth regarding long-term maintenance. I want to do this right, but I feel like I'm just pasting templates without fully understanding them.

I would love some wisdom from experienced maintainers on three specific things:

1. The CI/CD Workflow (Monorepo) My project is a monorepo (Backend/Frontend/CLI/MCP). I hacked together a GitHub Actions workflow that runs lint-and-build, but I don't know if it's efficient.

Should I be running separate workflows for each folder?
How do you handle versioning in a monorepo? (I'm currently bumping versions manually).
Is there a "gold standard" Action for testing a CLI tool?

2. Finding & Trusting Maintainers I am currently the sole developer. I know I can't do everything forever.

How do you identify "good" contributors who might become maintainers?
At what point do you give someone else write access to the repo?
How do I signal that I am open to mentorship/help without looking like I'm abandoning the project?

3. Blind Spots If anyone has a moment to glance at the repo structure, are there glaring security holes or anti-patterns I’m missing? I’ve enabled Dependabot and Branch Protection, but I don't know what I don't know.

Repo Link: https://github.com/binit2-1/Composter

Thanks in advance for helping a junior maintainer out!

https://redd.it/1prk9si
@r_opensource

What I Learned While Making an Open-Source Networking Library at IP Layer 2 in C Using Pcap

**I started SwiftNet about 8–9 months ago and have made over 200 commits. There were times I wanted to quit, but I'm glad I didn't. I've learned a lot about networking, debugging, and testing.**

Things I learned:

* Cross-platform testing is absolutely mandatory. Code that compiles and runs perfectly on macOS can break on Linux due to differences in headers, libraries, or behaviors. I run tests on multiple platforms (macOS and Linux) using GitHub Actions and only push changes when they pass.
* Thread safety in C is hard and time-consuming. Designing for thread safety from the start is crucial. Atomics (stdatomic.h) are often simpler, more performant, and less error-prone than mutexes. I've used atomics to create lock-free systems for several variables and built a custom memory allocator that pre-allocates memory per thread.
* Manual header construction is full of small mistakes. Working at Layer 2 (Ethernet frames) and Layer 3 (IP headers) requires calculating checksums yourself, handling endianness (htons/ntohs everywhere), and respecting MTU limits to avoid silent fragmentation issues.
* Error handling and debugging are a nightmare at this level. I've sometimes spent 10 hours on a single error caused by one small mistake. Use tools like Wireshark or Termshark to inspect packets.
* Always test with ASAN and TSAN, especially in multithreading. Even LLDB can't detect some issues that TSAN or ASAN can.

If you want to take a look at the source code, here’s the GitHub repo: [https://github.com/Morcules/SwiftNet](https://github.com/Morcules/SwiftNet)

https://redd.it/1prorom
@r_opensource