Open Source LLM Agents & Ecosystem Handbook — 60+ agent skeletons, tutorials, and ecosystem guides for building with LLMs

Hey all 👋

I’ve been working on the LLM Agents & Ecosystem Handbook, an open-source project designed to help developers explore and actually build with large language models.

What’s inside:
- 🛠 60+ agent skeletons (summarization, research, finance, healthcare, RAG pipelines, voice, games, MCP integrations…)
- 📚 Tutorials: RAG, Memory, Chat with X (PDFs/APIs/repos), Fine-tuning (LoRA, PEFT)
- ⚙ Ecosystem overview: training frameworks, local inference setups, LLMOps, interpretability
- 🔎 Evaluation toolbox: Promptfoo, DeepEval, RAGAs, Langfuse
- ⚡ Agent generator noscript to scaffold new projects quickly

It’s meant as a handbook (code + docs + ecosystem insights) — not just a link list.
The goal: help devs learn by building, and take agents from prototype to production.

👉 Repo link: https://github.com/oxbshw/LLM-Agents-Ecosystem-Handbook

Would love feedback from the OSS community — especially around contributions, documentation, and new agent ideas.


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

Released a self hostable observability tool for all your automations

Just published FlowMetr, a flexible lightweight observability tool for all workflows and pipelines out there on github.

Use it within your devops pipelines, source code or workflow tools like zapier, make or n8n

Can be used by everything capable of sending http requests.

Features:

- Metrics. How long are automations running?
- Logs. What was happening in run x yesterday?
- Alerts. Get notified when something breaks
- Reports you can share with your Team or your clients

Would be happy about feedback, stars, issues and contributions

Github here: https://github.com/FlowMetr/FlowMetr

https://redd.it/1nbfq74
@r_opensource

Open-source offline‑first multilingual translator based on ONNX converted MarianMT machine translation models.

https://github.com/harisnae/multilingual-translator-offline

Hosted on GitHub pages, no inference, runs entirely in the browser via @xenova/transformers. Loads the Ultra-light quantized model once, caches in IndexedDB, works without internet.
Currently supports 43 language pairs, I'm updating it to add more language pairs and list the English and non-English pairs separately. An important point, I I have currently set the max tokens to 256 (less than 200 words) so that it does not take long to translate.

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

Recovery Tool for Deleted APFS-snapshot and also for deleted files in general for macOS

Hi,

so unfortunately my APFS-snapshot got deleted (due to a restart initiated by a system software auto update) after half my home folder got deleted. Does anyone know of any open-source or at least free tooly/projecty to recover the deleted snapshot and also the deleted files of course. I generated an image, ran photorec, tried go-afps and afro (which both it seems are not functional at all). I am especially looking for pdfs, docx, txt, .py, (a lot of my academic work and also my academic library got deleted, also a lot of my programming projects) and also images (especially photos taken with an iPhone, so I guess they were .png?!). I would deeply appreciate it!

https://redd.it/1nbhq8j
@r_opensource

Pomodoro App with Widgets?

Ive searched far and wide for a pomodoro app with widgets that is open source and could not find any! Widgets that starts the timer directly from homescreen (android).

Also realised most open source dont have widgets at all.

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

Top 10 Open-Source AI Tools for Chat and Agents

Rasa: A leading open-source machine learning framework for building conversational AI. Rasa gives you complete control over your data and models, making it ideal for creating highly customized, context-aware bots for privacy-sensitive applications.
Botpress: This platform is an open-source conversational AI software that provides a visual builder and a powerful NLU engine. It's designed to be a balance of user-friendliness for non-technical users and flexibility for developers.
LangChain: A flexible framework for developing applications powered by language models. While not a chatbot platform itself, it's essential for orchestrating different components, allowing you to connect an LLM to external data sources and computation to build complex agents.
Weam.ai: This is a platform specifically designed to simplify the process of building and managing AI agents for teams. Now open source, Weam provides a structured framework to handle state management, tool integration, and conversational flows, allowing developers to focus on the agent's core logic. It's an excellent choice for teams looking for a collaborative, self-hosted solution.
Hugging Face Transformers: A foundational library that provides thousands of pre-trained models for a wide range of NLP tasks. It's a go-to for researchers and developers who want to fine-tune models or build custom, state-of-the-art AI agent architectures.
DeepPavlov: An open-source library for building conversational AI systems. It offers a comprehensive and flexible set of tools for creating production-ready conversational skills and multi-skill assistants, with pre-trained models and a user-friendly framework.
Microsoft Bot Framework: Microsoft's open-source offering for building, connecting, and deploying intelligent bots. It provides a code-driven platform with a wide range of tools and connectors for developers, offering fine-grained control over the chatbot experience.
Rasa X: While an extension of the core Rasa framework, Rasa X provides a web-based, low-code interface for reviewing and improving your AI assistants at scale. It's a key tool for teams to collaborate and manage the full lifecycle of their conversational AI.
CrewAI: An open-source framework designed for building collaborative, multi-agent workflows. It allows different AI agents with specialized roles to work together to tackle complex problems, making it a great choice for automation and problem-solving tasks.
ChatterBot: A simple, Python-based chatbot library that uses machine learning algorithms. It's designed to be a starting point for developers and is a great option for building basic conversational agents.

If you find any other tools please do let me know or we can discuss it here.

https://redd.it/1nbkc4u
@r_opensource

What is the go-to FOSS wiki hosting for videogames?

I want to start a wiki for a videogame that does not have it yet. After a previously common hosting, Fandom(.)com went through commercialization and enshittification, many videogame projects, like Minecraft, went the route of self-hosting. For it is a small game and I am not rich enough to rent a server and a domain for it, I cannot go this way.

When downsides of commercial aspects of GitHub were realized, Codeberg was created. Same with Mastodon/Bluesky to replace Twitter. Is there such a project to replace Fandom?

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

I can get my importan messages out of Telegram's mess thanks to this
https://github.com/Azizham66/WATele-Bridge

https://redd.it/1nbpm7k
@r_opensource

free, open-source file scanner
https://github.com/pompelmi/pompelmi

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

Alternatives to Loom (Windows)

Hi, I've used Loom but I'm not going to pay for it, but are there alternatives?

I'm also looking for software which records the zooming in and out of a webpage.

Thanks

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

I am working on a Open Source Desk Assistant
https://youtu.be/TJ5LkHuC5oM

https://redd.it/1nbuk1v
@r_opensource

Curious who the most active open source contributors are (globally & by country)? I built Gitista to find out - would love your thoughts!
https://www.gitista.com/

https://redd.it/1nbvfq3
@r_opensource

Postgres dominates the Metabase Community Data Stack Report 🐘

Just released our 2025 Data Stack Report with some interesting results from the database landscape.


PostgreSQL is absolutely crushing it, not only maintaining its lead as the top transactional database, but also emerging as the #1 choice for analytics storage.


Some standout findings:

PostgreSQL: 160 responses (nearly 3x more than MySQL at 56)
Traditional heavyweights like Oracle and SQL Server showing their age
27 people still say "I don't know" (we need to help them!)
MongoDB holding steady at 16 for NoSQL fans

Here is the full report for more insights about databases, data stacks, AI stuff, and what everyone's actually using these days.

https://redd.it/1nbqqx9
@r_opensource

Is there any third party tools enabling me to search among my Reddit saved posts



https://redd.it/1nbyv6y
@r_opensource

KeenWrite 3.6.3
https://keenwrite.com/

https://redd.it/1nbynp5
@r_opensource

I want to help you with testing

I'm a beginner qa manual/automation and I want to gain experience.

I'm familiar with Python and Selenium. I can also do manual work like writing test cases or bug reports.
I have no experience with unit tests and integration tests, but I really love to learn.

Let me know if I can help!

https://redd.it/1nbzi7i
@r_opensource

WatchdogFS: Rust CLI to watch files and alert on changes
https://github.com/gmaxferr/watchdogfs

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

Crafting a Distraction-Free E-Ink Dashboard to Outlive Google's Nest Hub

I hacked together an open-source e-ink display with TRMNL, pulling in custom APIs for weather, bus info, and personal health tracking.

The end result is energy-efficient and fully mine. https://youtu.be/MPm60wxAQKY?si=zLNTxU0rR0egOckZ

https://redd.it/1nc5as4
@r_opensource

dryrun - linux utility tool to perform dry run on your commands
https://github.com/tellsiddh/dryrun

https://redd.it/1nc9a4q
@r_opensource

I created a NinjaRipper 2 to OBJ converter noscript!

(09/08/2025)

Works for version 1.7.1 and below: https://github.com/Nazar-Okruzhko/NinjaRipper-NR-to-OBJ

If you don't know NinjaRipper is an awesome way to rip (a.k.a. extract) various 3D models from DirectX games.. hovewer it dumpes them into mysteriously-unkown .nr format... but I created a noscript for you so you can finally flex now!

All the magic will happen after F12 + this noscript!

https://redd.it/1nccx0g
@r_opensource

Why isn't it more common to create cross-platform and portable applications / software using web technologies like JS, HTML and CSS ?

I try to get rid of my reliance on proprietary ~~(Microsoft)~~ software with open source projects as much as I can. And regardless of the type of open-source software I'm looking for, I realized I have the following criteria that often come up :

OS compatibility : with Windows, Linux and MacOS
Device compatibility : with PC, smartphone and tablet
Out-of-the-box : No installation required, must be ready for use as is
Portability : can be used from a USB
No telemetry and no requirement to be connected to the internet
Self-contained dependencies to avoid complicated set-ups
Noob-friendly to download, execute and use by a tech-illiterate grandma

Optional criteria :

Syncing available across devices
Easy to change its source code to customize the software / web-app

I realize that pretty much all of these requirements are fulfilled with what would essentially be portable web-apps.

TiddlyWiki is one such example, it's a portable notebook that fits in one single HTML file (but I don't intend to do an implementation that extreme) and it works as intended.

Keep in mind that the alternatives for the type of software I'm looking for are not resource-intensive apps and are often light-weight :

Notes-taking markdown app (like Obsidian) / or text editor
E-book and manga reader that supports different file formats (PDF, EPUB, CBZ, etc.) and annotation
Very simple raster graphics editor like Paint
File converters
Meme maker

All of this being said, it circles back to my initial question :

Why isn't it more commonplace to use basic web technologies to create open-source projects for light-weight applications ? They seem to offer so much apparent advantages in addition to the fact that every OS and every device has a browser where these "apps" can run seamlessly.

So what gives?

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