Github Top Repositories
13.4K subscribers
1.58K photos
59 videos
10 files
2.16K links
Top GitHub repositories in one place πŸš€
Explore the best projects in programming, AI, data science, and more.
Download Telegram
Github Top Repositories
Photo
πŸ” Deep-diving into zai-org/GLM-5 β€” fresh off the trending list.

πŸ”— https://github.com/zai-org/GLM-5
πŸ“ GLM-5: From Vibe Coding to Agentic Engineering
──────────────────────────────

The GLM-5 series is a collection of open-source models designed for long-horizon tasks, complex systems engineering, and agentic engineering. The latest model, GLM-5.2, boasts a solid 1M-token context, advanced coding capabilities, and improved architecture, making it the strongest open-source model on standard coding benchmarks.

Key features include IndexShare, which reduces per-token FLOPs by 2.9Γ—, and speculative decoding, increasing acceptance length by up to 20%. The model is available for download on Hugging Face and ModelScope, with support for various frameworks like SGLang, vLLM, and Transformers.

Technical highlights of the GLM-5 series include DeepSeek Sparse Attention (DSA), asynchronous RL infrastructure, and the ability to control thinking budget through the reasoning_effort parameter. The model is suitable for researchers and developers working on long-horizon tasks, agentic engineering, and complex systems engineering.

To get started, you can download the model and follow the serve locally guide to deploy it with your preferred framework. Don't forget to cite the technical report if you find the model useful in your research.

In a nutshell, the GLM-5 series is a powerful tool for long-horizon tasks, and with its advanced features and support for various frameworks, it's an excellent choice for anyone looking to push the boundaries of AI research - so why not give it a try and see what you can achieve with GLM-5?

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
πŸ”₯ withastro/flue is trending β€” and it deserves your attention.

πŸ”— https://github.com/withastro/flue
πŸ“ The sandbox agent framework.
──────────────────────────────

Flue is a framework for building autonomous agents and powerful AI workflows with a programmable TypeScript harness. It allows you to create agents that can safely take action, maintain continuity, and connect to existing systems.

Key features include building agents that can keep context across conversations and events, running structured automations, providing a secure sandbox environment, and durable execution to preserve progress through failures and restarts.

Usage involves composing a harness with tools, skills, and instructions for the agent, and deploying it to a hosted runtime of choice.

Technical highlights include a built-in TypeScript harness, support for various deployment options like Node.js, Cloudflare Workers, and GitHub Actions, and integrations with tools like OpenTelemetry, Braintrust, and Sentry.

Audience includes developers looking to build the next generation of agents, with features like sandboxes, durable execution, and observability making it suitable for complex tasks.

createAgent is used to create an agent, with options for model, tools, skills, sandbox, and instructions.

export default createAgent(() => ({
model: 'anthropic/claude-sonnet-4-6',
tools: [...githubTools],
skills: [triage, verify],
sandbox: local(),
instructions,
}));


With Flue, you can unlock a new architecture for agents and build autonomous agents that can complete complex tasks - build agents that work for you, not the other way around.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
πŸ“Œ Spotted on GitHub Trending: n0-computer/iroh β€” let's break it down.

πŸ”— https://github.com/n0-computer/iroh
πŸ“ IP addresses break, dial keys instead. Modular networking stack in Rust.
──────────────────────────────

Iroh is a cutting-edge library that enables secure and efficient communication over the internet. At its core, iroh provides an API for dialing by public key, allowing users to establish fast and reliable connections with others, regardless of their location.

One of the key features of iroh is its ability to perform hole-punching, which enables direct connections between peers. If hole-punching fails, iroh can fall back to a network of public relay servers to ensure a stable connection.

Iroh is built on top of QUIC, a modern transport protocol that provides authenticated encryption, concurrent streams, and datagram transport. This allows iroh to offer a high level of security and performance.

To get started with iroh, users can install it using cargo add iroh and then use the provided examples to establish connections. Iroh also provides pre-existing protocols, such as iroh-blobs and iroh-gossip, which can be used to build custom applications.

The iroh repository is structured as a workspace of crates, including iroh, iroh-relay, and iroh-base. The project is licensed under the Apache License, Version 2.0, and the MIT license.

In summary, iroh is a powerful tool for building secure and efficient communication systems, and its robust feature set and flexible architecture make it an attractive choice for developers. With iroh, you can say goodbye to complicated networking code and hello to seamless, secure connections - it's a game-changer for distributed systems.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
❀1
⚑ obra/superpowers is making waves. Here's the full picture.

πŸ”— https://github.com/obra/superpowers
πŸ“ An agentic skills framework & software development methodology that works.
──────────────────────────────

Superpowers is a comprehensive software development methodology designed for coding agents, built on a set of composable skills and initial instructions. The purpose of Superpowers is to guide agents in developing high-quality code efficiently.

Key features include brainstorming, test-driven development, subagent-driven development, and code review. To use Superpowers, simply install it as a plugin in your preferred coding agent, such as Claude Code, Antigravity, or Codex App.

From a technical standpoint, Superpowers employs a skills library that includes testing, debugging, and collaboration skills. The system emphasizes test-driven development, systematic over ad-hoc approaches, and complexity reduction.

Superpowers is suitable for developers of all levels, from junior engineers to experienced professionals. The project is licensed under the MIT License and has a strong focus on community involvement, with a Discord channel and issue tracker available for support and feedback.

One-liner takeaway: Superpowers is a powerful tool that streamlines software development by guiding coding agents through a structured methodology, empowering developers to produce high-quality code efficiently.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
Github Top Repositories
Photo
🌟 penpot/penpot caught my eye on GitHub Trending today.

πŸ”— https://github.com/penpot/penpot
πŸ“ Penpot: The open-source design tool for design and code collaboration
──────────────────────────────

Penpot is an open-source design platform for teams that build digital products at scale, giving you full ownership of your design infrastructure. It's built on open source, supports self-hosting, and works with open standards like SVG, CSS, HTML, and JSON.

Key features include real-time collaboration, Design Tokens for consistency, and a powerful open API for automation and integrations. With CSS Grid and Flex Layout, teams can design responsive interfaces that behave like real code from the start.

Penpot is suitable for designers and developers, offering a fluid design-code process and real-time collaboration. The platform is deployment agnostic, allowing use in the cloud or deployment on your own servers.

To get started, you can use Penpot in the cloud or deploy it anywhere, and the community is always available to help. Whether you're a developer or designer, Penpot provides a single source of truth for design and development, making it a full-stack design platform for building scalable design systems.

Penpot's code-based approach and open API make it an ideal choice for teams that want to integrate design and development workflows. With its flexible and customizable nature, Penpot is perfect for teams of all sizes.

In short, Penpot empowers teams to own their design infrastructure, streamline collaboration, and build scalable design systems - all while maintaining full control and flexibility. Design, code, and collaborate - all in one place.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
Github Top Repositories
Photo
πŸ” Deep-diving into Kong/insomnia β€” fresh off the trending list.

πŸ”— https://github.com/Kong/insomnia
πŸ“ The open-source, cross-platform API client for GraphQL, REST, WebSockets, SSE and gRPC. With Cloud, Local and Git storage.
──────────────────────────────

Insomnia is an open-source, cross-platform API client for various protocols like GraphQL, REST, and more. With Insomnia, you can debug, design, test, mock, and build CI/CD pipelines for APIs. It offers features like a native OpenAPI editor, test suites, and collection runner. Insomnia supports multiple storage options, including Local Vault, Git Sync, and Cloud Sync, allowing users to choose how they want to store their projects.

The Insomnia API client is suitable for developers, testers, and teams who need to collaborate on API development. It's available for Mac, Windows, and Linux, and can be downloaded from the official website.

For developers who want to contribute to Insomnia, the project is structured as a monorepo and requires Node.js and Git to develop.

One-liner takeaway: Insomnia is the ultimate API client for developers, offering a wide range of features and flexibility to streamline API development and testing.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
πŸ“Œ Spotted on GitHub Trending: Lightricks/LTX-2 β€” let's break it down.

πŸ”— https://github.com/Lightricks/LTX-2
πŸ“ Official Python inference and LoRA trainer package for the LTX-2 audio–video generative model.
──────────────────────────────

Lightricks/LTX-2 is a groundbreaking DiT-based audio-video foundation model that packs all the core capabilities of modern video generation into one powerful package: synchronized audio and video, high fidelity, multiple performance modes, production-ready outputs, API access, and open access.

To get started, clone the repository and set up your environment with the required models, including the LTX-2.3 model checkpoint, spatial upscaler, temporal upscaler, distilled LoRA, and Gemma text encoder.

The model boasts several key features, such as available pipelines for production-quality text/image-to-video generation, video-to-video transformations, and audio-to-video generation. It also offers optimization tips for faster inference, including using distilled pipelines, enabling FP8 quantization, and installing attention optimizations.

Prompting for LTX-2 requires detailed, chronological descriptions of actions and scenes, including specific movements, appearances, camera angles, and environmental details. The model also supports automatic prompt enhancement via an `enhance_prompt` parameter.

The target audience for LTX-2 includes developers, researchers, and creators looking to generate high-quality videos using a versatile and powerful model.

Here's a quick start example:
git clone https://github.com/Lightricks/LTX-2.git
cd LTX-2
uv sync --frozen
source .venv/bin/activate


With LTX-2, the possibilities are endless - generate stunning videos with ease and take your creativity to the next level!

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
This media is not supported in your browser
VIEW IN TELEGRAM
LiteParse v2.1 β€” the fastest PDF β†’ Markdown parser in the world πŸš€

And at the same time, it's more accurate than any other open-source parser without using models (pymupdf4llm, opendataloader, pdf-inspector, markitdown) according to the results of three standardized benchmarks: olmOCR0-bench, opendataloader-bench, and ParseBench. πŸ“Š

Introducing LiteParse v2.1. The base version of LiteParse v2 was already the fastest document-to-text parser on the market, and in this release, Markdown support has been added. ✨

The project is completely open (Apache 2.0 license) and free. It's available via CLI, Rust, Node.js, Python, and WASM, and can also be installed as an agent skill in one click. πŸ› οΈ

More details: https://llamaindex.ai/blog/markdown-comes-to-liteparse

LiteParse repository: https://github.com/run-llama/liteparse

#LiteParse #PDF #Markdown #OpenSource #Rust #LLM

✨ Join Best TG Channels https://xn--r1a.website/addlist/0f6vfFbEMdAwODBk

⭐️ Join Our WhatsApp Channel https://whatsapp.com/channel/0029VaC7Weq29753hpcggW2A

πŸš€ Level up your AI & Data Science skills with HelloEncyclo β€” a growing all-in-one platform featuring hands-on courses in LLMs, Deep Learning, MLOps, Data Engineering, and more.
βœ… 13 courses live + 40+ coming soon
🎯 One access, lifetime updates
πŸ”‘ Use code: PRESALE-BOOK-WAVE-2GFG
πŸ‘‰ https://helloencyclo.com/?ref=HUSSEINSHEIKHO
❀1
Github Top Repositories
Photo
πŸ“Œ Spotted on GitHub Trending: palmier-io/palmier-pro β€” let's break it down.

πŸ”— https://github.com/palmier-io/palmier-pro
πŸ“ macOS video editor built for AI
──────────────────────────────

Palmier Pro is an open-source video editor built for AI, available for macOS 26 (Tahoe) on Apple Silicon. This Swift-native editor integrates generative AI features, allowing users to generate and edit videos together inside the timeline. Key features include built-in support for SOTA models like Seedance and Kling, as well as integration with agents like Claude, Codex, and Cursor via MCP server. To connect, users can follow the provided
claude mcp add --transport http palmier-pro http://127.0.0.1:19789/mcp
or similar commands for other agents. The editor is free to download and use, with optional subscription-based generative AI features. Join the community on Discord or follow on X for updates. With its unique AI-powered features and open-source nature, Palmier Pro is a game-changer for video editing - revolutionizing the way we create and edit videos, one timeline at a time.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
Github Top Repositories
Photo
πŸ’‘ penpot/penpot just hit the trending charts β€” here's why it matters.

πŸ”— https://github.com/penpot/penpot
πŸ“ Penpot: The open-source design tool for design and code collaboration
──────────────────────────────

Penpot is an open-source design platform for teams that build digital products at scale, giving you full ownership of your design infrastructure. With real-time collaboration, open standards like SVG, CSS, HTML, and JSON, and self-hosting capabilities, Penpot puts teams in complete control of their design environment.

Key features include best-in-class native Design Tokens, a powerful open API and plugin system, and multi-directional workflows between design and code. With CSS Grid and Flex Layout, teams can design responsive interfaces that behave like real code from the start.

Penpot is perfect for teams scaling their design systems, and its full-stack design platform helps build scalable design systems and integrated product development processes. Whether you're a designer or developer, Penpot's code-based approach makes designs readable by developers and AI.

Join the Penpot community to contribute, ask questions, and get inspired. You can participate in various ways, such as creating and sharing libraries and templates, reporting bugs, or contributing to the code.

Penpot is licensed under the Mozilla Public License, v. 2.0, and its source code is available on GitHub.

Takeaway: With Penpot, you can own your design infrastructure and build scalable design systems with ease.

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe
πŸ”₯ calesthio/OpenMontage is trending β€” and it deserves your attention.

πŸ”— https://github.com/calesthio/OpenMontage
πŸ“ World's first open-source, agentic video production system. 12 pipelines, 52 tools, 500+ agent skills. Turn your AI coding assistant into a full video production studio.
──────────────────────────────

Introducing OpenMontage, the first open-source, agentic video production system. This powerful tool allows you to create stunning videos by simply describing what you want in plain language. The agent handles research, scripting, asset generation, editing, and final composition, making it easy to produce high-quality videos.

Key features include the ability to start from a reference video, paste a YouTube link, and have the agent analyze the transcript, pacing, scenes, keyframes, and style to create a grounded production plan. You can also use Python 3.10+ and FFmpeg to install and run OpenMontage.

Technical highlights include the use of Remotion for React-based rendering, HyperFrames for HTML/CSS/GSAP rendering, and FFmpeg for post-production. The system also supports a range of API keys for additional tools and services, such as image and video gateways, free stock media, music, and voice services.

Audience includes anyone looking to create high-quality videos without extensive video production experience. Whether you're a marketer, educator, or simply a video enthusiast, OpenMontage is designed to be user-friendly and accessible.

In summary, OpenMontage is a revolutionary video production system that makes it easy to create stunning videos with minimal effort. With its powerful features and user-friendly interface, it's perfect for anyone looking to take their video production to the next level. Turn your ideas into professional-looking videos with OpenMontage - the future of video production is here!

──────────────────────────────
🧠 Channel: https://xn--r1a.website/GithubRe