Share your underrated GitHub projects

Hey everyone,

I spend a lot of time browsing GitHub Explore and love discovering interesting projects there, but I’m sure there are plenty of hidden gems that don’t get much visibility. If you have a repo you’re proud of, could you briefly explain what it does in 2–3 sentences and share the link? I’d love to check out and support some underrated projects.


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

Vibe Coded PRs?

So I recently opened my blockframe-rs project to MIT and that means anyone can contribute to the project. But I've just had my first pr and my god is it herrendous.

https://github.com/crushr3sist/blockframe-rs/pull/2

So the PR is +693 and -91.... I've advertised this project like 4 days ago. There's no way somone learnt my code, and found that my issues. But I think the puzzle gets so much easier to understand when you read the commit message which says "Add verbose personal anecdotes to all functions for line count inflation" and brings about fantastic changes to COMMENTS, and personally I think the dude's way ahead of me in terms of commenting, like check this out

before:

/// Tier 1 commit for files under 10MB. Uses RS(1,3) encoding where the whole file
/// is treated as a single data shard with 3 parity shards. File is padded to 64-byte
/// boundary (Reed-Solomon requirement), then 3 parity shards are generated.

After:

/// Commit tiny, like storing a small treasure in a safe deposit box. "Keep it secure," the banker says.
/// I'd read the file, pad to 64, generate parity, write files. "Protected!"
/// Committing tiny is like that – RS(1,3), create data and parity. "Safe deposit!"
/// There was this small item I kept losing, put it in a safe place. Peace of mind.
/// Life's about security, from treasures to files.

I mean this is pure poetry. But it doesn't end there, there's a lot more where that came from.

How do I keep this trash away from my project? I've not setup a PR guideline or done much to my Contributing markdown however, where i've shared that this project, I've explained this projects expectations with new contributors. But this is quite bad, I read about this in a few other posts, but I didn't think it would come to blockframe.

Vibe coding eh.

https://redd.it/1qb3qi3
@r_opensource

Built an open-source, subnoscription-free Geoguessr alternative
https://geoguesslite.com

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

Clipit - Desktop app for managing and sharing game clips
https://github.com/lowbit/clipit

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

CodeGraphContext: An MCP server plus a CLI tool that indexes local code into a graph database to provide context to AI assistants.
https://github.com/CodeGraphContext/CodeGraphContext

https://redd.it/1qb85xa
@r_opensource

Hybrid Search in PostgreSQL: The Missing Manual
https://www.paradedb.com/blog/hybrid-search-in-postgresql-the-missing-manual

https://redd.it/1qbg7hr
@r_opensource

built a contributor reputation checker to help with the AI slop problem

there's been a lot of discussion here about AI slop flooding open source. the curl situation, the 13k line OCaml PR, maintainers burning out on triage.                                                                                     

i want to help fix this. so i built something.                                                                                                           

SlopScore is a chrome extension that shows a reputation badge on github PRs.  

it catches the spray and pray pattern where someone submits mass low effort PRs across dozens of repos hoping something sticks.                                                                                                           

i started by listing scenarios that should obviously be green/yellow/red:                                   

  \- new account + 50 PRs in a week + 5% merge rate = red                        

  \- 200 merged PRs across mass repos over 3 years = green                       

  \- only self merges on own repos = suspicious                                                                                                                  

not sure i know i'm missing things. if you're a maintainer, what patterns do you actually see? what should i add? 

chrome store: https://chromewebstore.google.com/detail/slopscore/opcfceeomiijndecpjbkmmdpgpmclhli

source: https://github.com/hanzili/slopscore

demo: https://youtu.be/Y5pgfYhbzMA



https://redd.it/1qbck06
@r_opensource

Made CrawlChat open source

Being the admirer of open source community, I made CrawlChat open source!


I don't think there is any better open source Kapa ai alternative in market than CrawlChat. It lets you set up AI answering engine for your tech documentation and provides

A robust knowledge base management system
Keeps your knowledge base in sync with your sources
Embed the answering engine on Web, Discord, Slack, or as MCP
Observability layer to fine tune your docs

Super excited about this journey. Fork or star it here - https://github.com/crawlchat/crawlchat

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

How I used Svelte 5 to build a Vim-inspired game engine. Looking for TypeScript/Svelte contributors for zsweep

I built ZSweep because I wanted a Minesweeper experience that actually respects Vim grammar, not just a remapping of arrow keys to HJKL.

Using a mouse for grid navigation kept breaking my flow during short breaks, so I set out to design a game that is 100% keyboard-centric, where movement and actions feel composable in the same way Vim motions do.

Technical highlights:

Vim grammar engine: A custom state machine handles real Vim-like motions, including numeric prefixes (5j), search (/ + digit to jump to that number), and context-aware word motions (w / b) that skip revealed safe cells and land on unrevealed ones.

Svelte 5 & performance: Built with SvelteKit using the new Svelte 5 runes. Input latency was a hard constraint—if movement doesn’t feel as immediate as a terminal editor, the Vim feel completely falls apart.

Backend & stats: Powered by Supabase. I recently refactored the global stats dashboard from client-side counting to server-side aggregation (RPC) after hitting the classic 1,000-row select cap, which caused total time-swept stats to drift. Aggregates now reflect the full dataset correctly.

Security: Uses Row Level Security (RLS) so users can only modify their own game results, while keeping the global leaderboard public.

Why I’m sharing now: The project has crossed 1,000+ started games, and I’m looking for technical contributors to help polish the experience. I’ve opened GitHub issues around sound system customization and performance optimizations for larger boards (e.g. 30×16).

Play: https://zsweep.com

Source: https://github.com/oug-t/zsweep

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

Stop using MySQL in 2026, it is not true open source
https://optimizedbyotto.com/post/reasons-to-stop-using-mysql/

https://redd.it/1qbkd2f
@r_opensource

What is the real reason behind users' desire for OSS?
https://www.reddit.com/r/opensource/comments/1q9udjt/whats_the_one_proprietary_app_you_cant_find_a/

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

Open Source Screen Studio Alternative (Linux/Windows/macOS)

Link: https://github.com/tamnguyenvan/screenarc

https://redd.it/1qbk775
@r_opensource

I built a FreeMarker extension for Zed editor (with tree-sitter grammar)

Hey everyone! I just released a FreeMarker syntax highlighting extension for Zed editor, built with a custom tree-sitter grammar.

**What is this?** FreeMarker is a Java template engine that’s been around since 2000.
It’s still **widely used in enterprise systems**, Spring-based applications, and well-known projects like **Keycloak**, where it powers themes, login pages, and emails.

Despite that, modern editor support has been pretty lacking.

**Why Zed?** I recently switched to Zed and was frustrated by the lack of FreeMarker support. Since Zed uses tree-sitter for syntax highlighting, I decided to build a proper grammar from scratch rather than rely on regex hacks.

**A small disclaimer**:

I’m not an expert in grammar design or Rust, so the tree-sitter grammar is very much a learning-by-doing effort. That said, it’s been working well for real-world templates — and PRs, suggestions, and improvements are more than welcome.

**Features:**

* Full tree-sitter-based parsing (accurate even with complex nested structures)
* Both `<#...>` and `[#...]` syntax styles
* HTML injection for mixed templates
* All FreeMarker directives: conditionals, loops, macros, includes, built-ins
* Smart bracket matching and auto-closing

**Links:**

* GitHub: [https://github.com/debba/zed-freemarker](https://github.com/debba/zed-freemarker)
* Tree-sitter grammar: [https://github.com/debba/tree-sitter-freemarker](https://github.com/debba/tree-sitter-freemarker)

Built this mainly for my own projects, but figured others stuck maintaining FreeMarker templates might appreciate it. Let me know if you find any bugs or have feature suggestions!

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

Nametag - An open source personal relationships manager
https://nametag.one/

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

It's been a year since I published my first real FOSS project, here's what I learned

Hello everyone,

It has been a bit more than a year since I released Cyberbro.
It is my first real open source project.
The tool helps cybersecurity people, and anyone, to get information about IPs, domains, hashes and Chrome extension IDs.

At the beginning I was focused on stars.
The project now has around 500 stars, but I learned that stars are not the most important thing, even if this can help for visibility on Github.


What matters is to keep going, make releases, and accept that your code will not be perfect.
I used AI sometimes. If you use AI, please use an MIT license (this is my opinion, I'm very open to discuss about it).

I also created an MCP for the project, since it' AI times, right?
It works well, but I still prefer the real version.
Sometimes you do not need AI. You just need well presented data (we don't need to always use AI).

I tried to promote the project on Twitter and Mastodon.
It did not help much.
What helped more was creating a public demo and letting people try it.

If you build a web app, protect your server.
Do not forget this part!

For the repository, explain how people can report issues.
Be nice and don't be mean to newcomers.
I now have four maintainers. Two are active, and that is already a lot.
This is why open source is great.
One day you see that someone liked your project enough to use it and improve it.

Another nice thing is when people talk about your tool.
Sometimes I see posts on LinkedIn from security researchers who mention Cyberbro.
It feels good.
It also opens your world.
I met people from many countries.
The first person who reported issues was from Iran.
I got contributions from Germany, Texas, Sweden and more.

A small warning.
Be careful with monetization if your tool is niche. I was proposed to be sponsored.
It can look like a good idea, but people will expect results and you may lose your freedom.

I tried to write good documentation.
I take care of pull requests and issues.
I try to stay polite.
It's starting to pay off!

My advice is simple: your project will take time to grow.
There is a lot of noise today.
Do something that helps you first.
Then share it.

Thanks for reading!


I hope this post is useful.

https://redd.it/1qbrr0x
@r_opensource

Building a map-first CRM — looking for open-source-minded testers + contributors (repo soon)

Hi I’m building a lightweight CRM + sales funnel app with a map-based view (visualize leads/customers by location + pipeline/follow-ups). I’m trying to build it with an open-source mindset: transparent roadmap, public issue tracker, and a contributor-friendly setup.

Before I publish anything publicly, I’m looking for a small group of people who can help with:

* **Product feedback** (what’s missing, what’s confusing, what should be removed)
* **Self-hosting expectations** (docs, env setup, database choice, config)
* **Contributor experience** (what would make you *want* to contribute?)

If you’re interested, comment with your use case (field sales / local services / agency / etc.) and what you’d expect from an OSS-friendly CRM project. I’ll DM a link to the early build.

I’m not trying to promote — I genuinely want feedback from people who care about open collaboration and sustainable project practices.

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

Student contributor to CPython, NumPy, Pandas & Statsmodels looking to collaborate on open- source

Hii everyone

I’m a CS undergraduate who’s been actively contributing to open source over the past several months, mainly in the Python, machine learning, and scientific computing ecosystem.

I’ve contributed to projects including:

- CPython
- NumPy
- Pandas
- Statsmodels
- JAX-ml
- TensorFlow Quantum
- Apache Iceberg (Python)
- pgmpy
- sktime
- Data-8 Datascience
- Academy Software Foundation – OpenCue

I’m now looking to move beyond one off PRs and get involved more deeply, whether that’s --

"regularly contributing to a project"
"helping maintain a module"
or "joining a small OSS team that needs hands"

If you maintain or are part of a Python, ML, data, or infrastructure heavy project and could use extra hands feel free to reach out.

GitHub: https://github.com/Aniketsy

Thanks, and looking forward to connecting with folks here!



https://redd.it/1qbr669
@r_opensource

AI-startup's concepts are all same with our MIT-licensed OSS projects. Is this convergent evolution? or OSS etiquette violation?
https://typia.io/articles/ai-startup-concepts-are-all-same-with-our-mit-licensed-oss-projects.html

https://redd.it/1qbwt8c
@r_opensource

I built an open-source JSON viewer after getting tired of parsing nested stringified JSON at work

At work, I spend a lot of time digging through CloudWatch logs, and a recurring pain point was dealing with stringified JSON inside JSON… sometimes nested multiple levels deep.

You know the drill:

* JSON fields that are actually strings
* Those strings contain escaped JSON
* Sometimes *that* JSON contains another stringified JSON 🙃

Manually unescaping and reformatting this over and over got old very fast.

So I decided to build a small tool that would:

* Automatically detect and recursively parse stringified JSON
* Work entirely locally in the browser (no uploads, no backend)
* Make it easier to explore, query, and transform JSON while debugging

That tool eventually grew into **JSON Lens**, an open-source, keyboard-first JSON viewer & explorer.

What it does:

* Auto-parses deeply nested stringified JSON
* Code view + tree view
* JSONPath & jq-like queries
* Export to YAML / XML / TypeScript / NDJSON
* 100% local & privacy-friendly

It started as a “scratch my own itch” project, but I figured others dealing with logs, APIs, or configs might find it useful too.

Repo: [https://github.com/vakhariaheet/json-lens](https://github.com/vakhariaheet/json-lens)
Demo : [https://json.heetvakharia.in](https://json.heetvakharia.in)

Would love feedback, feature ideas, or critiques — especially from folks who deal with logs or observability tooling a lot.

Thanks for reading!

https://redd.it/1qbyux4
@r_opensource

Are there any opensource / modable TVs?

"Smart" TVs these days have like, the Youtube App, Amazon Prime, Netflix, etc. built in. These are usually littered with ads and paid features and have proprietary app stores so you can only have software they release on there. Not naming any manufacturer in particular, but I've owned several TVs that have done this. I'd like to have:

* Ability to install games, Brave Browser and other apps that would otherwise be restricted
* Parental controls so child doesn't grab remote, open browser and open bad websites
* The ability to install an app that can stream from a NAS or Plex/Jellyfin server.

https://redd.it/1qc14hp
@r_opensource

Use Maven dependency with AGPL

Hello everyone!

I am doing my final project for my IT degree. I need to make PDFs in my backend which is Spring Boot. I found there's this Maven dependency called iText, whichs seems to make the job I need quite easily.

The issue is I see some AGPL license, and I don't understand well how it affects a "private" (private during development, public after finished) project. How does affect it to my project? Could it put my final project or university in danger?

Thanks for your time and answers :3

https://redd.it/1qc25oa
@r_opensource