📌 Spotted on GitHub Trending: Leonxlnx/taste-skill — let's break it down.

🔗 https://github.com/Leonxlnx/taste-skill
📝 Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop
──────────────────────────────

Taste Skill is an innovative GitHub repository designed to enhance the capabilities of AI agents in building premium frontends. The repo offers a range of portable agent skills that focus on improving layout, typography, motion, and spacing, resulting in more sophisticated and polished UIs.

Key features include adjustable dials for design variance, motion intensity, and visual density, allowing for customized outputs. The skills are framework-agnostic and compatible with major coding agents like Codex, Cursor, and Claude Code.

To get started, simply install the desired skill using npx skills add https://github.com/Leonxlnx/taste-skill, then use the installed skill in your agent conversations. For image-generation skills, pair them with ChatGPT Images or similar generators to produce design images.

The repo is suitable for developers and designers looking to elevate their frontend builds with AI-driven design skills. With its unique approach to anti-slop design and extensive research backing, Taste Skill is a valuable tool for anyone seeking to create high-end interfaces.

In a nutshell, Taste Skill is a game-changer for AI-powered frontend development - revolutionizing the way we design and build premium UIs, one skill at a time.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

🌟 cursor/plugins caught my eye on GitHub Trending today.

🔗 https://github.com/cursor/plugins
📝 Cursor plugin specification and official plugins
──────────────────────────────

The Cursor plugins repository offers a collection of official plugins for various developer tools, frameworks, and SaaS products. Each plugin is a standalone directory with its own manifest, making it easy to manage and maintain.

Key features include continual learning for incremental memory updates, thermos for deep security audits, and create-plugin for scaffolding new plugins.

To use these plugins, simply navigate to the desired plugin directory and follow the instructions in the README.md file.

From a technical standpoint, the repository uses a marketplace.json file to list all available plugins, and each plugin has its own plugin.json manifest.

These plugins are perfect for developers and dev teams looking to streamline their workflow and improve productivity.

In short, the Cursor plugins repository is a game-changer for development teams - it's like having a superpower in your coding toolkit.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

📌 Spotted on GitHub Trending: run-llama/liteparse — let's break it down.

🔗 https://github.com/run-llama/liteparse
📝 A fast, helpful, and open-source document parser
──────────────────────────────

LiteParse is a fast and lightweight, open-source PDF parsing tool that delivers high-quality spatial text parsing with bounding boxes. It runs locally on your machine, without relying on proprietary features or cloud dependencies.

Key features of LiteParse include fast text parsing using PDFium, a flexible OCR system with built-in Tesseract and support for HTTP servers, screenshot generation, and multiple output formats like JSON and text. It also supports multi-language use from Rust, Node.js/TypeScript, Python, or the browser (WASM) and is multi-platform, compatible with Linux, macOS, and Windows.

To install LiteParse, you can use your preferred package manager. For example, you can install it via npm for Node.js/TypeScript, pip for Python, or cargo for Rust.

The CLI usage is straightforward. You can parse files using the lit parse command, generate screenshots with lit screenshot, and perform batch parsing with lit batch-parse.

Technical highlights include automatic conversion of various document formats to PDF before parsing, support for office documents via LibreOffice, and image formats via ImageMagick.

Audience: LiteParse is suitable for developers and users who need fast and accurate PDF parsing capabilities, especially those working with large volumes of documents or requiring precise text positioning information.

In summary, LiteParse is a powerful, user-friendly tool that provides fast and accurate PDF parsing, making it an excellent choice for anyone looking to extract valuable information from their documents - Parse your documents, unleash the power of your data.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

🚀 Meet galilai-group/stable-worldmodel: a gem from today's GitHub trending list.

🔗 https://github.com/galilai-group/stable-worldmodel
📝 A platform for reproducible world model research and evaluation
──────────────────────────────

The stable-worldmodel repository provides a unified platform for world model research and evaluation. It offers a single interface for data collection, training, and evaluation with model-predictive control across various environments. The repository includes reference implementations of common baselines and planning solvers, allowing researchers to focus on their core contributions.

The library supports multiple data formats, including lance, hdf5, folder, video, and lerobot, and provides tools for format conversion and benchmarking. It also features a large suite of environments, including those from the DeepMind Control Suite, Gymnasium, and OGBench, with factors of variation for evaluating zero-shot generalization.

To get started, users can install the library via PyPI and follow the quick start guide to collect data, train a world model, and evaluate it using model-predictive control. The repository is in active development, with APIs subject to change between minor versions.

The stable-worldmodel library is designed for researchers and developers working on world models, and its unified interface and reference implementations make it an ideal choice for those looking to advance the state-of-the-art in this field.
With stable-worldmodel, the world is your model, and the possibilities are endless.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

🎯 byoungd/English-level-up-tips landed on trending. Worth a proper look.

🔗 https://github.com/byoungd/English-level-up-tips
📝 An advanced guide to learn English which might benefit you a lot 🎉 . 离谱的英语学习指南/英语学习教程/英语学习/学英语
──────────────────────────────

The English-level-up-tips GitHub repository is a comprehensive guide to learning English, dedicated to the author's past love, W. This project aims to provide a detailed and advanced guide to help individuals improve their English skills.

The guide covers various aspects of the English language, including understanding, vocabulary, listening, reading, speaking, and writing. It also features a chapter on AI and its application in language learning.

The guide is suitable for anyone looking to improve their English skills, from beginners to advanced learners. It's a valuable resource for those who want to learn English in a natural and enjoyable way.

Key features of the guide include its comprehensive coverage of the English language, its use of AI in language learning, and its focus on making learning English a fun and rewarding experience.

The guide is available for online reading on various platforms, including GitHub Pages, GitBook, and Zhihu.

In short, the English-level-up-tips guide is a must-have resource for anyone looking to improve their English skills, and with dedication and practice, you can master the English language and unlock a world of new opportunities.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe

🎯 Biohub/esm landed on trending. Worth a proper look.

🔗 https://github.com/Biohub/esm
📝 No description.
──────────────────────────────

The Biohub/esm GitHub repository presents a groundbreaking world model of protein biology, leveraging the latest advancements in Evolutionary Scale Modeling (ESM). This comprehensive system comprises three primary components: ESMC, ESMFold2, and ESM Atlas.

ESMC is a state-of-the-art protein language model trained on billions of protein sequences, allowing it to learn and represent the rules of protein biology. ESMFold2 builds upon ESMC and is a state-of-the-art structure prediction model that can predict high-resolution, all-atom 3D protein structures directly from amino acid sequences. The ESM Atlas is a vast map of 6.8 billion proteins, organized according to the internal world model of ESMC, enabling the prediction of over one billion structures.

These tools can be utilized through the Biohub Platform or by running the models locally with Hugging Face. The repository provides extensive documentation, including tutorials and preprints, to facilitate understanding and usage. The target audience includes researchers, scientists, and developers interested in protein biology, structure prediction, and world modeling.

Here is a simple example of running ESMC locally:

import torch
from transformers import AutoModelForMaskedLM, AutoTokenizer

# example GFP sequence
sequences = ["MSKGEELFTGVVPILVELDGDVNGHKFSVSGEGEGDATYGKLTLKFICTTGKLPVPWPTLVTTFSYGVQCFSRYPDHMKQHDFFKSAMPEGYVQERTIFFKDDGNYKTRAEVKFEGDTLVNRIELKGIDFKEDGNILGHKLEYNYNSHNVYIMADKQKNGIKVNFKIRHNIEDGSVQLADHYQQNTPIGDGPVLLPDNHYLSTQSALSKDPNEKRDHMVLLEFVTAAGITHGMDELYK"]

model = AutoModelForMaskedLM.from_pretrained(
    "Biohub/ESMC-6B",
    device_map="auto",
).eval()
tokenizer = AutoTokenizer.from_pretrained("Biohub/ESMC-6B")

inputs = tokenizer(sequences, return_tensors="pt", padding=True)
inputs = {k: v.to(model.device) for k, v in inputs.items()}
with torch.inference_mode():
    output = model(**inputs)

To get started, simply install the esm library and import the necessary modules. With Biohub/esm, unlock the power of world modeling in protein biology and discover new frontiers in protein structure prediction and design. The future of protein biology is here, and it's being modeled with unprecedented accuracy.

──────────────────────────────
🧠 Channel: https://t.me/GithubRe