What makes a good first issue?

I maintain an open-source Python program for recording data from software-defined radios called *Spectre*. It's reasonably niche, so one of our prime focuses has been to make it as accessible as possible for new developers.

It's hosted on GitHub, and I've recently been brainstorming ideas for good first issues. For me, these would be straightforward and have a clearly defined, small scope. For example, I created an issue which concerns removing some functions which were made redundant after a recent refactor.

I'd be keen to hear from the community what you think makes a good first issue? For maintainers: which issues do you label that are likely to be picked up by new contributors? For contributors: when exploring a new repository, what qualities do you look for in an issue before deciding to make your first contribution?

https://redd.it/1p6kgds
@r_opensource

An open, error‑driven learning framework that could become cross‑subject infrastructure

I’ve started an open source project that’s intentionally small in scope but ambitious in its implications: an adaptive, error‑driven learning framework that’s meant to be reused and remixed across subjects.

Right now it runs on top of an LLM and is wired for learning Python. But the core of the project is not “Python tutorials” – it’s the *infrastructure* around learning:

* A strict “study mode” where the AI only reports line numbers and error types (syntax / structure / logic / input), no full solutions by default.
* A hard limit of **two new concepts per lesson**, with automatic splitting into sub‑lessons if you exceed that (cognitive load baked into the design).
* A goal/lesson structure (G##, L##, W##, T##) with a command/skill tracker, progress tracking, error history and a learning log.
* Built‑in metacognitive reflection at the end of each session that feeds back into the plan.

All of this is encoded in config and JSON/Markdown files, so it’s transparent, hackable, and auditable. The LLM is “just” the execution engine; the pedagogy lives in the repository.

The potential, and the reason I think this belongs in open source rather than as a closed product:

* **Cross‑subject reuse**: You can swap out “Python commands” for grammar rules, math techniques, physics concepts, etc. If we build multiple subject templates (Esperanto, physics, statistics, …), we get data on how far one didactic skeleton can stretch.
* **Shared improvements**: Every time someone refines the error taxonomy, the reflection prompts, or the session structure, that improvement can immediately benefit all subjects using the same core.
* **Transparent AI behavior**: Instead of “black box tutoring”, the rules the AI must follow (no full solutions, error types only, max 2 new concepts, required reflection) are defined in code, versioned, and reviewable.
* **Privacy‑friendly by design**: Personal logs/progress are kept out of the template; what we share are only structures and rules, not user data.

Best case scenario in an open source context:
Over time, this could evolve into a kind of “didactic kernel” – a shared, community‑maintained engine for error‑driven, reflective learning, with domain‑specific templates contributed by different people (languages, STEM, humanities, etc.). The more subjects we plug in, the more we learn about what generalizes and what doesn’t.

Repo (currently a template for Python, with English docs and no personal data):
[https://github.com/Tobzu/-adaptive-learning-system-](https://github.com/Tobzu/-adaptive-learning-system-)

If you’re interested in:

* stress‑testing the didactic assumptions,
* adapting it for your own subject, or
* helping to turn this into a more general open framework for AI‑assisted learning,

I’d love feedback, criticism, and contributions.

https://redd.it/1p6nbzu
@r_opensource

KeenWrite survey

Hi there! I'm seeking directions to take KeenWrite, my free, open-source, cross-platform, desktop Markdown editor. Any feedback you're willing to offer would be most helpful:

https://www.surveymonkey.com/r/WGDGG79

https://redd.it/1p6m58c
@r_opensource

openDesk 1.10. Enhanced security architecture
https://www.opendesk.eu/en/blog/opendesk-1-10

https://redd.it/1p6qolb
@r_opensource

What They Don't Tell You About Maintaining an Open Source Project
https://andrej.sh/blog/maintaining-open-source-project/

https://redd.it/1p6q7mr
@r_opensource

I'm building a C-based json processing language... in json.

https://github.com/flintwinters/unnoscriptd-jisp

I would like to build a community around it, and there is a discord link in the readme.

I'm implementing the language in C using the yyjson library which you can find here: https://github.com/ibireme/yyjson it is the fastest json parser available.

The language works by just looping over a json array in a json object to modify that object's own structure. This means a program in the language is completely self contained. You could stop a program in the middle of executing and copy its current state as a simple json object and email it to someone and they could continue where you left off.

I have already added the option to store each operation's residual value as a JSON patch, which means you can actually go backwards while debugging a program.

I have a bunch more tasks planned, check out the todo on the github.

https://github.com/flintwinters/unnoscriptd-jisp

https://redd.it/1p6tfs7
@r_opensource

Pixeli - The CLI Tool for Creating Beautiful Image Grids and Mosaics
https://github.com/pakdad-mousavi/pixeli

https://redd.it/1p6yen6
@r_opensource

Launching Open Source Voice AI
https://rapida.ai/opensource?ref=reddit_opensource

https://redd.it/1p709ho
@r_opensource

Is x265 open source?

I'm a bit confused on whether x265 is actually open source. I'm aware that H.265 is not open source and had complex licensing/royalty annoyances, but then apparently x265 is void of this. How is this so (if this is true)?

https://redd.it/1p71bqj
@r_opensource

LF opensource project

Hi.
i whant to get involved in opensource community and would like to find a project that is more like an opensource integration platform like Jitterbit or something alike.
suggestions on how to get involved in opensource?

https://redd.it/1p71ssn
@r_opensource

Collected 100+ Open Source projects offering jobs.
https://open-source-jobs.com

https://redd.it/1p73cl6
@r_opensource

Wingfoil - the ultra-low latency data streaming framework built in Rust

Hi,

we've just launched Wingfoil, an open source, ultra-low latency data streaming framework built in Rust, and we're looking for feedback and / or contributors.

You can find Wingfoil on crates and Github.

Wingfoil is highly scalable and simple to use - it has a user-friendly API and easily integrates with existing tools. Wingfoil is built in Rust, but has browser bindings in Python and integrates with Tokio to simplify the setup of asynchronous I/O adapters.

Let us know what you think.


https://redd.it/1p73tyf
@r_opensource

What are the Indian-origin Open Source shopping cart systems like Magento or WooCommerce?

What Indian-origin open-source shopping cart systems are available that offer capabilities similar to global platforms like Magento or WooCommerce? I’m looking for solutions developed in India that provide flexibility, strong customization options, marketplace support, and community-driven development for building scalable eCommerce stores.

https://redd.it/1p75x49
@r_opensource

[Pre-release] We're opensourcing our entire AI Middleware stack, as the other existing ones weren't fitting the bill or had too many issues to work around.
https://github.com/rootflo/wavefront

https://redd.it/1p76t4v
@r_opensource

Released: Torrra v2 - a fast, modern terminal torrent search & download tool

Hey everyone!
I’ve just shipped **Torrra v2**, a big upgrade to my TUI torrent search/download tool built with Python & Textual.

**What’s new in v2:**

* Faster UI + smoother navigation
* Improved search experience
* Better multi-torrent downloads
* Cleaner indexer integration
* Polished layout + quality-of-life tweaks

I cannot post the full intro video here, so please check this out,
Full video: [https://youtu.be/NzE9XagFBsY](https://youtu.be/NzE9XagFBsY)

Torrra lets you connect to your own indexer (Jackett/Prowlarr), browse results, and download either via libtorrent or your external client; all from a nice terminal interface.

If you want to try it or check out the code:
**GitHub:** [github.com/stabldev/torrra](http://github.com/stabldev/torrra)

Feedback, ideas, and PRs are welcome!

https://redd.it/1p77q5p
@r_opensource

What’s an example of a big open-source *app*?

We’ve all seen plenty of open-source libraries and smaller utilities.

Those codebases are quite different from production apps that have all the things:

* billing
* feature flags
* CI flows
* schemas & migrations
* APIs
* component libraries
* e2e tests
* cli
* doc site
* shared utilities
* etc

I think the [Excalidraw](https://github.com/excalidraw/excalidraw), [Cal](https://github.com/calcom/cal.com), and [Posthog](https://github.com/PostHog/posthog) repos are well-structured, for example.

But there’s gotta be more good ones.

Any repos you'd recommend I check out?

(Trying to build some good mental models as my [open-source calendar](https://github.com/SwitchbackTech/compass) app grows to avoid some pain)

https://redd.it/1p77ol0
@r_opensource

I built a CLI tool to backup my GitHub repos (issues, wikis, releases) because I was paranoid about losing data

Hello . I've been working on a Python CLI tool called [Farmore](https://github.com/miztizm/farmore) to help automate backing up GitHub data. I realized that while `git clone` is great for code, it doesn't really save the "metadata" like issues, releases, and wikis that I use for project management.

I built this to be a simple "set and forget" tool. It runs concurrently, so it’s pretty fast even if you have a lot of repos. It handles:

* Cloning/Updating repos
* Exporting issues and wikis
* Downloading releases

It’s open source, and I’d love to hear if there are any edge cases I missed or features that would make it more useful for your workflows.

https://redd.it/1p79phc
@r_opensource

Curious how others handle open-source office tools in their workflow

I’ve been trying to move more of my daily tools into the open-source ecosystem, especially for document editing and collaboration. I experimented with ONLYOFFICE recently inside a self-hosted setup, mostly to see how well it fits into an open-source workflow, and it’s been smoother than I expected.

Before I commit to it fully, I wanted to ask the community:
What open-source office or document tools have you found reliable for everyday work?

I'm less interested in “best of all time” lists and more in real experiences, what actually holds up over months of use, how updates affect stability, and which tools integrate well into larger self-hosted or open-source environments.

Would love to hear what’s been working (or not working) for you.

https://redd.it/1p79cmh
@r_opensource

AgentSudo - Permission system for AI agents.

I’m excited to share **AgentSudo**, a small open-source permission system for AI agents.

# What My Project Does

AgentSudo lets you **assign scoped permissions** to AI agents and protect Python functions using a decorator — just like the `sudo` command in Unix.

Example:

from agentsudo import Agent, sudo

support_bot = Agent(
name="SupportBot",
scopes=["read:orders", "write:refunds"]
)

analytics_bot = Agent(
name="AnalyticsBot",
scopes=["read:orders"]
)

(scope="write:refunds")
def process_refund(order_id, amount):
print(f"Refunded ${amount} for {order_id}")

# Support bot can process refunds
with support_bot.start_session():
process_refund("order_123", 50) # ✅ Allowed

# Analytics bot cannot
with analytics_bot.start_session():
process_refund("order_456", 25) # ❌ PermissionDeniedError

The idea is to prevent real damage when LLM-based agents hallucinate or call unsafe tools.

# Target Audience

AgentSudo is for:

* Developers using AI agents in **production** (customer support bots, automation, internal tools)
* People working with **LangChain, AutoGen, LlamaIndex**, or custom multi-agent frameworks
* Anyone who needs **least-privilege** execution for AI
* Researchers exploring **AI safety / tool use** in practical applications

It works in any Python project that calls functions “on behalf” of an agent.

# Comparison to Existing Alternatives

Most existing AI frameworks (LangChain, AutoGen, semantic tool-use wrappers):

* Provide *tool calling* but **not real permission boundaries**
* Rely on LLM instructions like “don’t delete the database,” which aren't reliable
* Use a **single API key** for all agents
* Have no built-in audit trail or scope enforcement

AgentSudo is:

* **Framework-agnostic** (wraps normal Python functions)
* **Super lightweight** (no infra, no cloud, no lock-in)
* **Declarative** — you define scopes once per agent
* Inspired by real security patterns like OAuth scopes & sudo privileges

# Links

* **GitHub:** [https://github.com/xywa23/agentsudo](https://github.com/xywa23/agentsudo)
* **PyPI:** [https://pypi.org/project/agentsudo](https://pypi.org/project/agentsudo)
* **Product Hunt launch:** [https://www.producthunt.com/products/agentsudo](https://www.producthunt.com/products/agentsudo)

It’s MIT-licensed — feedback, criticism, PRs, or ideas are very welcome.

Thanks! 🙌[](https://www.reddit.com/submit/?source_id=t3_1p7ckwi)

https://redd.it/1p7drg1
@r_opensource

Journiv Self Hosted Journal: This Thanksgiving, give your family the gift of memories that last forever

Hello everyone!

First of all, thanks a lot for the [amazing response](https://www.reddit.com/r/opensource/comments/1oni77v/journiv_011beta_is_out_a_selfhosted_privacyfirst/) and interest in Journiv. We have [hundreds of stars](https://github.com/journiv/journiv-app/stargazers), thousands of [docker pull](https://hub.docker.com/r/swalabtech/journiv-app) and many many [feature request](https://github.com/journiv/journiv-app/issues) (and bugs reports) on Github in just two weeks (sleepless two weeks for me :)).

[Journiv](https://journiv.com) v0.1.0-beta.8 is out and in it I have added the most requested features. [Github](https://github.com/journiv/journiv-app)

Journiv is available on [Unraid Community Apps](https://github.com/JPDVM2014/unraid-templates/blob/main/journiv.xml).

**Highlights:**

* OIDC support (now pretty stable)
* In app [one click export-import](https://www.youtube.com/watch?v=rQRpQbyExMU) with history. So you always have your memories safe and backed up even if you don't want to deal with docker backups
* Role Based Access Control for user management.
* Many quality of life features and bug fixes.
* Read the release notes [here](https://github.com/journiv/journiv-app/releases/tag/v0.1.0-beta.8)

Journiv began as a deeply personal project, a way for me to capture memories, reflections, and the stories behind thousands of photos and videos of my fast-growing kids. What started as a tool for my own parenting journey has grown into something that fills a real gap in the self-hosting community.

If you’re curious, you can read the full story behind Journiv [here](https://journiv.com/blog/the-story-behind-journiv).

I’m grateful that Journiv is now helping others preserve their memories as well.

**The Journey Ahead**

Journiv is in active development, with a fully functional backend, a web frontend, and mobile apps launching soon. It is self-hosted, and designed to be your companion for decades.

Journiv is being built because our memories deserve to be ours, forever.

**So this Thanksgiving, give your family the gift of memories that last forever!**

https://redd.it/1p7hf7p
@r_opensource

Timeplus Proton 3.0: Up to 7x Performance Gains in Pipeline Processing
https://www.timeplus.com/post/proton-3-0

https://redd.it/1p7ibdi
@r_opensource