Github Top Repositories

💡 twentyhq/twenty just hit the trending charts — here's why it matters. 🔗 https://github.com/twentyhq/twenty 📝 The open alternative to Salesforce, designed for AI. ────────────────────────────── The Twenty GitHub repository is home to an open-source CRM that empowers technical teams to create custom solutions tailored to their unique business needs. With Twenty, you can build, ship, and version your CRM like any other application in your stack. To get started, you can either sign up for a cloud-based workspace, scaffold a new app using the npx create-twenty-app my-app command, or self-host Twenty on your own infrastructure using Docker Compose. Twenty provides a range of features, including objects, views, workflows, and agents, which can be extended as code. The platform also supports version control, allowing you to track changes and updates to your CRM. On the technical side, Twenty is built using a stack that includes TypeScript, Nx, NestJS, PostgreSQL, and React. Whether you're a developer looking to customize your CRM or an organization seeking a flexible and scalable solution, Twenty has something to offer. The takeaway: Twenty is the CRM that you can build, shape, and evolve like your own application - giving you the freedom to create the perfect tool for your business needs. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🎯 mukul975/Anthropic-Cybersecurity-Skills landed on trending. Worth a proper look. 🔗 https://github.com/mukul975/Anthropic-Cybersecurity-Skills 📝 754 structured cybersecurity skills for AI agents · Mapped to 5 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND & NIST AI RMF · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 26 security domains · Apache 2.0 ────────────────────────────── Anthropic Cybersecurity Skills is an open-source library of 754 production-grade cybersecurity skills, covering 26 security domains and mapped to five industry frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, MITRE D3FEND, and NIST AI RMF. This library is designed to give AI agents the security skills of a senior analyst, enabling them to execute structured decision-making workflows and follow practitioner playbooks. The skills are encoded in a consistent directory structure, with each skill having a SKILL.md file containing YAML frontmatter and Markdown body sections. The YAML frontmatter provides metadata, such as the skill name, description, and tags, while the Markdown body sections outline when to use the skill, prerequisites, step-by-step workflow, and verification procedures. Key features of this library include: * 754 production-grade cybersecurity skills * 26 security domains covered, including cloud security, threat hunting, and digital forensics * 5 framework mappings for unified cross-framework coverage * Compatible with 26+ AI platforms, including Claude Code, GitHub Copilot, and OpenAI Codex CLI To get started, users can clone the repository or use npx skills add mukul975/Anthropic-Cybersecurity-Skills. The library is designed for AI agents to use, but security professionals and developers can also benefit from it. Takeaway: With Anthropic Cybersecurity Skills, AI agents can now think like senior security analysts, and that changes everything. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🔥 shiyu-coder/Kronos is trending — and it deserves your attention. 🔗 https://github.com/shiyu-coder/Kronos 📝 Kronos: A Foundation Model for the Language of Financial Markets ────────────────────────────── Kronos is the first open-source foundation model for financial candlesticks, trained on data from over 45 global exchanges. It's a decoder-only model, pre-trained to handle the unique, high-noise characteristics of financial data. The model uses a novel two-stage framework, first quantizing continuous, multi-dimensional K-line data into hierarchical discrete tokens, and then pre-training a large, autoregressive Transformer on these tokens. Key Features: - Pre-trained on data from 45 global exchanges - Novel two-stage framework for handling financial data - Decoder-only model for autoregressive forecasting - Supports multiple model sizes for different computational needs The model can be used for forecasting by loading a pre-trained Kronos model and its corresponding tokenizer from the Hugging Face Hub, and then using the `KronosPredictor` class to handle data preprocessing, normalization, prediction, and inverse normalization. from model import Kronos, KronosTokenizer, KronosPredictor The model supports batch prediction for efficient processing of multiple time series, and also provides a pipeline for fine-tuning on your own datasets. Technical Highlights: - Novel two-stage framework for handling financial data - Supports multiple model sizes for different computational needs - Batch prediction for efficient processing of multiple time series The model is suitable for anyone interested in financial forecasting, including quantitative traders, researchers, and students. Audience: - Quantitative traders - Researchers - Students In summary, Kronos is a powerful tool for financial forecasting, and its open-source nature makes it accessible to anyone. With its novel framework and support for batch prediction, it's an ideal choice for anyone looking to improve their financial forecasting capabilities. Get ready to forecast like a pro with Kronos - the future of financial forecasting is here! ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

📌 Spotted on GitHub Trending: p-e-w/heretic — let's break it down. 🔗 https://github.com/p-e-w/heretic 📝 Fully automatic censorship removal for language models ────────────────────────────── The Heretic project is a game-changer in the world of language models, offering fully automatic censorship removal without the need for expensive post-training. By combining directional ablation with a TPE-based parameter optimizer powered by Optuna, Heretic enables users to decensor language models with ease. The process is completely automatic and doesn't require configuration, although Heretic provides various parameters for greater control. Users can run Heretic using a simple command, replacing the model name with their desired model. For example:

heretic Qwen/Qwen3-4B-Instruct-2507

Heretic supports most dense models, including multimodal and MoE architectures, and has been used to create over 3000 models. The community has well-received models generated with Heretic, praising their ability to provide uncensored responses without destroying the model's intelligence. In addition to its primary function, Heretic offers research features for exploring model internals, including generating plots of residual vectors and printing details about residual geometry. These features can be accessed by installing Heretic with the optional research extra. In short, Heretic is a powerful tool that's democratizing access to uncensored language models - and that's a big deal! ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🔥 Leonxlnx/taste-skill is trending — and it deserves your attention. 🔗 https://github.com/Leonxlnx/taste-skill 📝 Taste-Skill - gives your AI good taste. stops the AI from generating boring, generic slop ────────────────────────────── The Taste Skill project is an anti-slop frontend framework designed for AI agents, focusing on creating premium interfaces with strong layout, typography, motion, and spacing. This framework offers a range of agent skills that can be easily installed using the npx skills add command. The skills include various design and image-generation capabilities, such as design-taste-frontend, imagegen-frontend-web, and brandkit. Key features of Taste Skill include: - Adjustable dials for design variance, motion intensity, and visual density - Support for multiple coding agents, including ChatGPT and Codex - Framework-agnostic design rules - Image-generation skills for creating reference images To get started with Taste Skill, simply run the installation command:

npm install https://github.com/Leonxlnx/taste-skill

Taste Skill is ideal for developers and designers looking to enhance their AI-built interfaces with a more premium and polished look. In short, Taste Skill is a game-changer for anyone looking to take their AI-designed interfaces to the next level - it's time to add some taste to your skill. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🚀 Meet anthropics/knowledge-work-plugins: a gem from today's GitHub trending list. 🔗 https://github.com/anthropics/knowledge-work-plugins 📝 Open source repository of plugins primarily intended for knowledge workers to use in Claude Cowork ────────────────────────────── The anthropics/knowledge-work-plugins GitHub repository offers a collection of plugins designed to enhance the capabilities of Claude, a workplace assistant. These plugins cater to various roles and teams, providing a strong foundation for tasks such as productivity, sales, customer support, and more. Key features include: - Plugin Marketplace: 11 open-sourced plugins for different job functions - Customization: Edit plugins to fit your company's tools and workflows - Connectors: Integrate with various tools like Slack, Notion, and Jira To get started, users can install plugins directly from Cowork or via

claude plugin install

commands. The structure of each plugin includes a plugin.json manifest, .mcp.json for tool connections, commands for explicit actions, and skills for domain knowledge. These plugins become more valuable when customized for your company's specific needs, allowing you to swap connectors, add company context, adjust workflows, and build new plugins. The takeaway: with the anthropics/knowledge-work-plugins, you can transform Claude into a tailored specialist for your team, streamlining workflows and amplifying productivity. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

💡 affaan-m/ECC just hit the trending charts — here's why it matters. 🔗 https://github.com/affaan-m/ECC 📝 The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. ────────────────────────────── The ECC (harness-native operator system) is a complete system for agentic work, born out of an Anthropic hackathon win. It's not just about configurations, but rather a robust framework that includes skills, instincts, memory optimization, continuous learning, security scanning, and research-first development. With 182K+ stars, 28K+ forks, and 170+ contributors, ECC supports 12+ language ecosystems and works across various AI agent harnesses like Claude Code, Codex, and Cursor. The system is designed to be production-ready, with features like agent and skill management, hook and rule configurations, and memory optimization techniques. ECC also includes a range of tools and features for security, research, and development, such as AgentShield and cross-harness architecture. To get started with ECC, users can follow the Shorthand Guide or the Longform Guide, which provide detailed information on setup, foundations, and philosophy. The Security Guide is also available for users who want to learn more about attack vectors, sandboxing, and sanitization. ECC is suited for developers, researchers, and operators who want to build and deploy AI-powered agents and skills. With its comprehensive framework and wide range of features, ECC is an ideal choice for those looking to create robust and scalable agentic systems. Takeaway: ECC is a powerful and flexible operator system that empowers users to build, deploy, and manage AI-powered agents and skills with ease, and its harness-native approach makes it a game-changer in the world of agentic work. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🔍 Deep-diving into hardikpandya/stop-slop — fresh off the trending list. 🔗 https://github.com/hardikpandya/stop-slop 📝 A skill file for removing AI tells from prose ────────────────────────────── The Stop Slop GitHub repo is all about helping you refine your writing by removing predictable patterns that are common in AI-generated content. This skill is designed to be used with large language models (LLMs) like Claude, teaching them to catch and remove AI tells such as clichéd phrases, structures, and rhythms. To get started, you can add this skill to your Claude code, upload the core instructions and reference files to your project knowledge, or use API calls to include the skill in your system prompt. The repo includes a list of banned phrases and structural clichés to avoid, as well as sentence-level rules to improve your writing. The repo also includes a scoring system to help you evaluate your writing, with dimensions such as Directness, Rhythm, Trust, Authenticity, and Density. The Stop Slop skill is perfect for anyone looking to improve their writing and make it sound more human. Use it to refine your content and take your writing to the next level - after all, good writing is rewriting. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🚀 Meet Lum1104/Understand-Anything: a gem from today's GitHub trending list. 🔗 https://github.com/Lum1104/Understand-Anything 📝 Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more. ────────────────────────────── The Understand Anything GitHub repo is a powerful tool that turns any codebase, knowledge base, or documentation into an interactive knowledge graph. This graph can be explored, searched, and questioned, making it easier to understand complex systems. The tool works with various platforms, including Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more. Key features include: - Explore the structural graph: Navigate the codebase as an interactive knowledge graph. - Understand business logic: Visualize how the code maps to real business processes. - Analyze knowledge bases: Turn a wiki into a navigable graph of interconnected ideas. To use Understand Anything, simply install the plugin, analyze your codebase, and explore the dashboard. The tool also supports multi-platform installation and can be shared with teammates by committing the graph to the repository. Technical highlights include a tree-sitter + LLM hybrid approach that combines static analysis and LLMs for efficient and accurate results. The tool is suitable for developers, project managers, and anyone looking to gain a deeper understanding of complex systems. In summary, Understand Anything is a versatile tool that simplifies the process of understanding complex codebases and knowledge bases. With its interactive graph, business logic visualization, and knowledge base analysis, it's an essential tool for anyone looking to cut through the noise and understand the big picture. The ultimate takeaway: Understand Anything helps you stop reading code blind and start seeing the big picture. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

🚀 Meet st-tech/ppf-contact-solver: a gem from today's GitHub trending list. 🔗 https://github.com/st-tech/ppf-contact-solver 📝 A contact solver for physics-based simulations involving 👚 shells, 🪵 solids and 🪢 rods. ────────────────────────────── Introducing ZOZO's Contact Solver, a cutting-edge physics-based simulation tool for shells, solids, and rods developed by ZOZO, Inc., Japan's largest fashion e-commerce company. This robust and scalable solver resolves contacts penetration-free, handling extreme cases with over 180M contacts. Key features include a cache-efficient design, finite element method for deformables, and symbolic force jacobians, all running on the GPU for massively parallel performance. The solver is Apache 2.0 licensed, allowing for commercial and proprietary use. To get started, users can choose between a Windows native executable or a Docker environment, with optional Blender add-on and JupyterLab integration for a seamless simulation experience. Technical highlights include a Python API for customization and a GPU-accelerated design for fast performance. This project is perfect for developers, researchers, and engineers looking for a reliable and efficient contact solver. In a nutshell, ZOZO's Contact Solver is a powerful tool that makes complex physics-based simulations easy and accessible - giving you the power to simulate anything, anywhere. ────────────────────────────── 🧠 Channel: https://t.me/GithubRe

📌 Spotted on GitHub Trending: thedotmack/claude-mem — let's break it down. 🔗 https://github.com/thedotmack/claude-mem 📝 Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More ────────────────────────────── Claude-Mem is a persistent memory compression system designed for Claude Code. It seamlessly preserves context across sessions by automatically capturing tool usage observations, generating semantic summaries, and making them available to future sessions. Key features include persistent memory, progressive disclosure, and skill-based search. The system is easy to install with a single command: npx claude-mem install. Claude-Mem is suitable for developers and power users looking to enhance their productivity with Claude Code. The system provides a web viewer UI and supports natural language queries for searching memory. The project is well-documented, with a dedicated documentation website that covers installation, usage, and configuration. In a nutshell, Claude-Mem is a game-changer for Claude Code users, and you can supercharge your productivity with it - give it a try! ────────────────────────────── 🧠 Channel: https://t.me/GithubRe