Guidance needed ! New to open source.

Hey folks, I am an undergrad who wants to start with open source. I am not much into the dev side. I mostly work with building ML models working on kaggle. How should I start with open source particularly in field of AI/ML?
Also I have heard about gsoc being a good opportunity, any help on that will also be great.

https://redd.it/1oyfis9
@r_opensource

No-Trust Protocol for Backtesting Systematic Trading Algorithms

TL;DR: Backtesting trading algorithms lacks a transparent, reproducible standard. Results can’t be reliably verified. Could a no-trust, open protocol help?


Backtesting is the foundation of systematic trading algorithms — yet there’s still no open, verifiable standard for how backtests should be recorded, structured, reproduced, or audited. Everyone seems to be using their own JSON/CSV formats. You can usually read another person’s backtest output, but you can’t reliably verify it.

I’m thinking about a no-trust protocol: a specification defining how backtests should be logged, hashed, documented, and reproduced. It’s not a product or a platform, just an open protocol anyone can implement.

Key ideas could include:

fixed, open schemas for inputs and outputs

cryptographic consistency checks

required metadata for full reproducibility

deterministic execution guidelines

fully open-source reference tools

complete auditability, zero-trust assumptions


A decentralized, peer-to-peer implementation could ensure backtest data remains publicly verifiable while avoiding central control. The protocol would need to remain neutral and non-commercial to preserve its integrity.

I’m just a beginner exploring this idea, so this is more a thought than a proposal. Does anyone know if something like this already exists?

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

Is there a Topaz Video Alternative for Linux

Hello! Im looking for a Topaz alternative for Linux. I already found a similar post but it's 3 years old and I think things probably changed and it was about Topaz Pictures: https://www.reddit.com/r/opensource/comments/uwm4vx/comment/iluf9lu/?utm\_source=share&utm\_medium=web3x&utm\_name=web3xcss&utm\_term=1&utm\_content=share\_button

So are there any video alternatives for linux?

Thank you!

https://redd.it/1oyi3kt
@r_opensource

The new Aider-CE fork of Aider AI Assistant is now official - Hard Fork!

Aider was and is one of the best Open Source AI coding assistants. Unfortunately the original maintainer has ignored the project he started for months now - to the point where there are now 1000+ unresolved issues and 200+ un-merged PR's on the Aider GitHub!!

A few contributors have been working on an unofficial fork for a few months now, hoping to get the changes merged in later. But recently the original creator of Aider started deleting any mention of "aider-ce" from the Aider discord. We have had to do a hard fork.

Blog post: https://www.circusscientist.com/2025/11/16/the-new-aider-ce-fork-of-aider-ai-assistant-is-now-official/

https://redd.it/1oyr3z4
@r_opensource

Big milestone reached - arkA building end to end

Big milestone reached — arkA Protocol is now fully building end-to-end!

In 48 hours we took a brand-new repo and:
• fixed dozens of npm / ESM / Rollup dependency issues
• rebuilt the entire CI/CD system (linting, schema validation, builds)
• repaired Markdown formatting across all docs
• cleaned and validated every schema & example file
• modernized the codebase for Node 18+ + ESM
• restored the reference client build

This gives arkA its first fully reproducible build pipeline.

arkA is NOT “another YouTube clone.”
It’s a content metadata protocol that any app can use to describe, index, and discover video in a fully open ecosystem.

Looking for curious devs who want to help shape an open alternative to opaque recommendation algorithms and locked-down creator platforms.

Repo here → https://github.com/baconpantsuppercut/arkA

https://redd.it/1oyrwa6
@r_opensource

A Fully Open Source Peer-to-Peer Social Protocol That Anyone Can Build Apps or Clients On Top Of
https://github.com/plebbit/plebbit-js

https://redd.it/1oysk8h
@r_opensource

EHTML — Extended HTML for Real Apps. Sharing it in case it helps someone.

Hi everyone! I’ve been working on a project called EHTML, an HTML-first approach to building dynamic pages using mostly HTML. It lets you handle things like templating, loops, conditions, data loading, reusable components, and nested forms — all without a build step or heavy JavaScript setup.

I originally built it to simplify my own workflow for small apps and prototypes, but I figured others who prefer lightweight or no-build approaches might find it useful too. It runs entirely in the browser using native ES modules and custom elements, so there’s no bundler or complex tooling involved.

If you enjoy working close to the browser or like experimenting with minimalistic web development, you might find it interesting. Just sharing in case it helps someone or sparks ideas. Cheers!


Link: https://e-html.org/

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

I got tired of js frameworks… so I wrote my own in Kotlin

Over​‍​‌‍​‍‌ a year ago I had a plan to create a web framework - because I was fed up with js/ts ecosystems and I wanted a simple, predictable, and fully Kotlin-based solution.

After a lot of the times trying and refactoring, the project is finally at a point where I think it’s ready to share.

### What it is

A minimal full-stack Kotlin web framework with:

- API routing

- HTML routing (with dynamic rendering)

- a very small mental model

- no large dependency chain

- simple setup → fast to understand

- still flexible enough for real projects

### Why I built it

Ktor and Spring may be good, but they are large ones. What they need is time to be learned, and they bring a lot of patterns that you are forced to adapt to.

I wanted to have something small, see-through, and that is easy to be understood - and also I wanted to know how internally the frameworks work instead of the usual relying-on-magic.

### If that sounds interesting, you can try it

GitHub: https://github.com/Jadiefication/Void

Jitpack: https://jitpack.io/#Jadiefication/Void

I’m not stopping until it’s perfect, and I would be super happy to have feedback from other Kotlin developers that would like to have a small but powerful alternative in the ​‍​‌‍​‍‌ecosystem.

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

looking for contributors - python library

🎨 CM-Colors: Making web accessibility easier - Looking for contributors!

[What it is](https://github.com/comfort-mode-toolkit/cm-colors/)**:** A Python library that automatically improves color contrast for WCAG compliance while preserving visual aesthetics (using perceptual color science).

**Current state:** Core library works great, now expanding with:

* 🐛 Parser improvements (good first issues available!)
* 🖥️ CLI tool for processing CSS files
* 📊 Batch processing and reporting features

**Looking for:**

* Python developers (beginner to advanced)
* CLI/UX enthusiasts
* Accessibility advocates
* Anyone interested in color science!

**Repo:** [github.com/comfort-mode-toolkit/cm-colors](http://github.com/comfort-mode-toolkit/cm-colors)

**Good first issues:** We have well-documented starter tasks with pseudocode

Check out issue #26 for a great entry point! 🚀

I know it can feel scary to make your first contribution, here are some resources to help you get started:
\- [Contribution Guide with clear steps to get started](https://github.com/comfort-mode-toolkit/cm-colors?tab=contributing-ov-file#readme)
\- [Codebase tour of cm-colors](https://gist.github.com/lalithaar/286ce89bc1d4816b3d6c862380ddc9d6)
\- [How to code when you have chosen an issue](https://comfort-mode-toolkit.github.io/wiki/code/code-how-to/)
\- [Acessibility basics in plain language and why it matters](https://comfort-mode-toolkit.github.io/wiki/accessibility-basics/)

Feel free to let me know if you have any questions

https://redd.it/1oz4wmg
@r_opensource

I Probably Made The Most Simplest Build System Ever Existed 😹 & It's Battle Tested FR
https://github.com/darkyboys/cook

https://redd.it/1oz9mzw
@r_opensource

web based e-mail-client

Hello everyone,

I’m looking for a web-based email client, as the noscript says. What I mean by that is that I want something like Thunderbird, where I can manage multiple mailboxes, identities, and calendars from different email providers.

The reason is that I have many email addresses for different purposes, and I want to bundle them across all my devices.

Thanks a lot in advance.

https://redd.it/1ozan7o
@r_opensource

Beginner looking for paid open-source issues (even small bounties) — where should I start?

Hi everyone,
I’m a fresher trying to get into open-source, but I also want to earn a little while I learn. I’ve already tried programs like Outreachy and GSoC but wasn’t selected.

Now I’m looking for something simpler:
👉 Open-source projects that offer small paid issues/bounties
👉 Beginner-friendly places to contribute and get paid as I grow

If you know any platforms, projects, or communities that regularly post paid issues even $5–$20 bounties. I’d really appreciate your suggestions.

Thanks!

https://redd.it/1oz9988
@r_opensource

I open-sourced MemLayer, a Python library that adds persistent long-term memory to LLM applications

# What My Project Does

MemLayer is an open-source Python library designed to give LLM-based applications persistent, long-term memory.
LLMs normally operate statelessly. Every interaction starts fresh, with no continuity between calls.

MemLayer adds a small but useful layer on top of existing LLM clients:

* it captures important information from conversations,
* stores it locally and persistently (vector + optional graph memory),
* and retrieves the relevant context on later calls so the model can answer with continuity.

The idea is to enable more consistent and contextual behavior without rewriting your application or adopting a large framework.

# Target Audience

MemLayer is meant for:

* developers building LLM features in Python
* anyone who wants stateful behavior without maintaining their own memory backend
* researchers exploring memory architectures for LLMs
* open-source projects that want a standalone memory component
* people who prefer local, dependency-minimal tooling

It works fully offline, with any LLM provider or local model, and requires no external services.

# Comparison With Existing Alternatives

MemLayer differs from larger frameworks in a few ways:

* Focused: It only handles memory, not orchestration, agents, or pipelines.
* Pure Python: Small codebase, easy to read, modify, or extend.
* Local-first: No required cloud APIs; memory is stored entirely on disk.
* Structured memory: Uses semantic vector search and optional graph storage.
* Noise-aware: Includes an optional ML-based gate to avoid saving irrelevant content.

The goal is to provide a simple, transparent component rather than a full ecosystem.



Happy to get feedback, suggestions, or contributions.
If you’re interested in the design or want to help shape future features, I’m all ears.

GitHub: [https://github.com/divagr18/memlayer](https://github.com/divagr18/memlayer)
PyPI: pip install memlayer

https://redd.it/1ozc5br
@r_opensource

OpenLinux — new from-scratch Linux distribution looking for contributors (boot, libc, toolchain, docs)
https://github.com/openlinux-src/src

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

Working on an opensource IaaS platform for Hyper-V

Hello!

I’ve been working on OpenHVX, an open-source project aiming to bring a full IaaS layer to Microsoft Hyper-V, similar to what Proxmox or Harvester do, but built specifically for Hyper-V environments.
The goal is to make it easier to manage hosts, VMs, networks, quotas, and multiple tenants through a modern web UI and API, without relying on SCVMM or any other vendor-locked-in tooling, though it can still be complementary in some cases.

It’s still under active development and not production-ready yet, but the foundation is solid and progressing quickly.
I’m open to feedback, testing, and ideas. PRs are very welcome if you’d like to jump in.

If you’re curious:
Website: https://openhvx.org
Docs: https://openhvx.org/docs
Github: https://github.com/openhvx

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

update on making your colors accessible without losing the brand

Hello guys, few months back I shared about the open source library I was working on called cm-colors

this post is more of something that happened which made me really happy than anything

So there was this friend in my class who was working on a website and chose a really pretty theme, yk those aesthetic one and he was really satistfied with his work

He ran it through the wcag color contrast checkers and found that some pairs ( like those used on buttons etc ) didnt pass AA :((

He was dicussing about how disappointed he was ( the website was to suprise his gf, so he used her fav colors ) when we were hanging out and we tried to put it through cm-colors ( I was not quite sure since even tho I coded the library to ensure it keeps the design intent, because the before and after looking the exact same almost )

But then I used devtools in chrome to see the contrast has indeed changed and there wasn't a bug in the library lol

This was the original usecase I built the library for, choosing a palette that looekd really good but wasnt accessible, like it wasnt totally invisible but it still didnt cross AA quite

But overtime I felt like I was the only one with that usecase lol, so it was pretty nice to see someone else had the same use too :>

Inspired by his work, I created a demo and ran the before after through https://www.whocanuse.com/ and it indeed worked yayyyy - kudos to the team behind whocaseuse so I know I wasn't deluding

That said, one of my classmate started working on the literature review for how color contrast affects people with vestibular needs - it makes me so happy to see my classmates slowly becoming aware of learning to build with accessibility and how it's about most of us in different times

I am not sure if this sound's salesy or anything, As much as I am happy if the library spreads and more people start making accessible websites, I am not sharing the links here for any purpose other than setting context for the incident - so you dont have to click any links unless you want to :>

This also made me feel so grateful for all the work wcag, and all the a11y community efforts into making a more accessible web

https://redd.it/1ozgdd5
@r_opensource

An ai native open source Git worktree manager CLI that works with all your ai coding agents
https://github.com/coderabbitai/git-worktree-runner

https://redd.it/1ozeyo3
@r_opensource

Introducing the OpenNDA

[Lawyer Here but also a techie\]

This is something I have been working for a while. Am launching it into the comments phase.

OpenNDA is an open, Creative-Commons-style Non-Disclosure Agreement. Affix the notice, the recipient opens the media, and acceptance is complete. Includes modular codes for jurisdiction, term, confidentiality, and commercialization limits. Simple, automatic, and universally usable.

A Creative-Commons-style NDA.

No signatures.

No DocuSign.

No “please sign before we can talk.”

Just attach the notice.

They open the file/email.

The NDA is automatically in force.

Meet OpenNDA.

Simple. Universal. Free.

Find Out More at : https://github.com/thatlawyerfellow/OpenNDA and see if you'd like to help standardise it.[Lawyer Here but also a techie\]

This is something I have been working for a while. Am launching it into the comments phase.

OpenNDA is an open, Creative-Commons-style Non-Disclosure Agreement. Affix the notice, the recipient opens the media, and acceptance is complete. Includes modular codes for jurisdiction, term, confidentiality, and commercialization limits. Simple, automatic, and universally usable.

A Creative-Commons-style NDA.

No signatures.

No DocuSign.

No “please sign before we can talk.”

Just attach the notice.

They open the file/email.

The NDA is automatically in force.

Meet OpenNDA.

Simple. Universal. Free.

Find Out More at : https://github.com/thatlawyerfellow/OpenNDA and see if you'd like to help standardise it.

https://redd.it/1ozj479
@r_opensource

OpenMicrofrontends Specification

Hi all,

I and the maintainers of OpenMicrofrontends are pleased to announce the first release of our microfrontend specification. Now, microfrontends have no clear definition and the term is applied rather broadly to different technologies.

We aim to provide an open standard for defining/describing microfrontends by drawing from our experience in the field in developing such systems. Please, if you are interested, check out our Official Page, which provides a variety of examples! We are happy for any feedback, suggestions and questions!

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

Open source tools for PR summaries?

I’ve been looking for open-source tools that can summarize pull requests automatically. Most of what I find are paid products or closed systems that plug into GitHub or GitLab.

What I’m hoping for some of you to helo with me is something lightweight that can generate human-readable summaries from PR diffs (ideally per commit or per file) and maybe post a comment or summary block. Even better if it can run on-prem or inside CI without depending on a hosted API.

I’ve seen CodeRabbit and Bito do this nicely, but I’d rather use (or contribute to) something open. Does anything out there come close? Or are people here just rolling their own with local LLMs or huggingface pipelines?

Would love examples or repos. Mainly want something that helps reviewers keep up without needing to read 30-file diffs line by line.

Thanks all!

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

QuicShare – Fast, secure, peer-to-peer file sharing (built with .NET + Avalonia)

Hi Friends!

I just released QuicShare, a simple and lightweight peer-to-peer file sharing app. It’s designed to make sending files between two devices super easy — no cloud, no central servers, just direct transfers.

Repo link: GitHub – QuicShare

# Why it’s great

Easy to use – just create a room, share the code, and start sending files.
Direct transfers – files go straight from your device to your peer’s device.
Secure – end-to-end encryption with QUIC + mutual TLS.
Unlimited file size – send large files without worrying about limits.
Cross-platform – works on Windows 11 (x64 & ARM64) and Linux.
Privacy-friendly – the signaling server only helps peers connect; your files never leave your devices.

# How it works

1. One peer creates a room.
2. Share the room code with your peer.
3. Both peers connect directly, and transfers happen securely and instantly.

This project is all about making file sharing quick, private, and effortless. Feedback is super welcome! And if you find it useful, a star on the repo would mean a lot.

GitHub – QuicShare

https://redd.it/1ozn98l
@r_opensource