New tools, libraries, and projects gaining traction on GitHub
69 stories in the last 7 days
Moonshot AI releases Kimi K3 open weights and infrastructure after shaking up the frontier model race
Kimi K3 is an open-weight frontier model that nearly matches top Western systems on standard benchmarks. Moonshot AI has published the mo…
jasoseo-plugin — 한국 취업 자기소개서를 기업조사→초안→글자수→리뷰→윤문→검증 8단계로 완성하는 Claude Code 플러그인
jasoseo-plugin is an open-source Claude Code plugin for automating Korean job application cover letters. The tool guides developers throu…
Why China is giving away its best AI models
Moonshot AI releases open-weight Kimi K3 model to challenge US dominance. The high-performing LLM delivers competitive capabilities at a …
Beyond RAG: Task-aware knowledge compression for enterprise AI on AWS
TAKC is an open-source framework for task-aware knowledge compression on AWS infrastructure. Traditional RAG struggles with analytical ta…
council-lab — Local-first, human-participatory four-agent AI deliberation workspace
Council-lab is a local-first multi-agent workspace for collaborative AI deliberation. Built with Python, Next.js, and LangGraph, the tool…
AI YC interview with Gstack agents
Gstack is an open-source toolkit of AI voice bot agents that join Google Meet calls to provide in-persona YC interview feedback. The proj…
old-coder — An old coder's strategy for the agent era: don't read the code — make it run the gauntlet. Evidence-first development sk
old-coder is an evidence-first development toolkit designed to help coding agents test code rather than reading it. Inspired by classic s…
Why Silicon Valley is rallying behind open-weight AI - PRESS Insider
Open-weight AI models are reshaping how Silicon Valley builds and deploys machine learning systems. Major tech companies and startups ali…
stormy-cookbook — Open-source cookbook for the Stormy Social Data API and MCP server (Model Context Protocol) — one REST API for the TikTo
Stormy Cookbook is an open-source guide and MCP server for multi-platform social data APIs. The repository provides a single REST interfa…
openclaude-improved — runs anywhere. uses anything
OpenClaude-Improved is a versatile, TypeScript-based open-source coding agent that runs anywhere and integrates with multiple tools. The …
ru-marketplace-mcp — MCP-серверы для российских маркетплейсов: Wildberries, Ozon, Яндекс Маркет, Детский мир и сравнение цен по всем сразу. Т
ru-marketplace-mcp is a Model Context Protocol server enabling Claude to query Russian e-commerce data. The Python-based tool provides re…
codex-chatgpt-web — Use ChatGPT Pro as a native model in the Codex app — with context, compaction, streaming, images, and optional MCP tools
Codex-chatgpt-web is an open-source macOS tool integrating ChatGPT Pro natively into the Codex app. It supports features like full contex…
llm-engineering-platform — A production-oriented LLM engineering platform with OpenAI-compatible serving, streaming, observability, evaluation, rep
llm-engineering-platform is an open-source LLMops toolkit for production-grade AI applications. It offers OpenAI-compatible serving, stre…
synapse-enforcer-leet — Thinking-native compiler for LEET Language Understanding item sets — killer-item design grammar, verification separation
Synapse-enforcer-leet is a thinking-native compiler for LEET Language Understanding item sets. The project features killer-item design gr…
data-driven-ai-guide — Teach Data Engineers How to Think AI.
Data-driven-ai-guide is a curated repository helping data engineers build reliable AI foundations. It provides practical resources and ar…
AI Code Prep v2
AI Code Prep v2 is a Rust-powered CLI tool that formats large repositories for LLM context windows. The rewrite delivers massive speed im…
Braidkeep
Braidkeep is an open-source CLI tool that predicts and prevents merge conflicts for parallel AI coding agents. The single binary utility …
deer-workflow — An open-source Dynamic Workflow runtime that keeps orchestration in TypeScript and delegates semantic work to replaceabl
deer-workflow is an open-source TypeScript runtime for orchestrating dynamic AI agent workflows. It keeps core logic in TypeScript while …
forge-os — The open-source control plane for AI agents — skill routing, context governance, trustworthy execution, evidence, securi
Forge-os is an open-source control plane for managing AI agent execution and orchestration. The JavaScript-based platform provides develo…
llmwiki-harness
LLMwiki-harness is a Python-based utility for managing personal knowledge bases using Markdown and Obsidian. It integrates with AI coding…
avenoxskills — Production agent skills for Claude Code, Cursor, and any SKILL.md harness — Codex fleets, video pipeline, monorepo revie
avenoxskills is a collection of production-ready agent skills for popular AI coding harnesses. The repository provides modular capabiliti…
ViralSim — 让一群 AI 用户先刷到你的笔记：模拟小红书图文的停留、阅读与互动，并用真实发布数据持续校准爆款概率。
ViralSim is an open-source multi-agent simulation tool for predicting social media content performance. It simulates user engagement, dwe…
ADE
ADE is an open-source multi-platform client that syncs all your AI coding agents across devices. It unifies existing coding subscriptions…
Heard
Heard is a macOS voice layer bringing spatial audio summaries to agentic workflows. It integrates with Claude Code, Codex, and Cursor to …
Capsomnia
Capsomnia is a free, open-source utility that keeps MacBooks awake by remapping the Caps Lock key. The tool prevents macOS from sleeping …
Nvidia, Microsoft and other tech giants back open-source AI models
Major tech companies are backing open-source AI models to bypass restrictive guardrails on proprietary systems. Hugging Face recently use…
Drift — Desktop coding agent using the OpenCode agent engine
Drift is a desktop coding agent built on the OpenCode agent engine. Written in TypeScript and Rust using Tauri and SolidJS, the tool inte…
German AI consortium releases Soofi S, an open 30B model that tops benchmarks in both English and German
Soofi S 3.0 is a bilingual 30-parameter open-source model tainted by training data contamination. The German consortium behind the model …
autodev-studio — Autonomous multi-agent SDLC harness: describe a feature in plain English and AI agents scope, code, test, review, and op
Autodev-studio is an open-source autonomous multi-agent software development harness. Developers can describe a feature in plain English,…
Grok-UI — Local-first live command center for Grok Build — runtime, ACP control, session history, and Git changes.
Grok-UI is a local-first dashboard for managing Grok Build local runtimes and sessions. The TypeScript-based tool provides developers wit…
capybara — Terminal trace debugger for AI agents.
Capybara is a Go-based terminal trace debugger built specifically for AI agents. The open-source CLI tool leverages OpenTelemetry to help…
OpenWorker
OpenWorker is a local-first desktop AI agent that automates tasks in everyday applications. The open-source tool runs directly on user ha…
scientific-illustrator — Editable scientific figures in PowerPoint and draw.io via Codex/MCP with Designer-Drawer-Reviewer-Corrector quality gate
scientific-illustrator is an open-source tool generating editable scientific figures for PowerPoint and draw.io. It leverages Codex and t…
VinvAI — Your agent says it's done. Vinv says prove it. Real traces + live code graph + closed-loop verify, served to your agent
VinvAI is an open-source observability tool that verifies AI coding agent completions using live code graphs. The tool provides real exec…
Poolside's Laguna S 2.1 is a small open-weight coding model that punches well above its size
Laguna S 2.1 is an open-weight coding model designed for autonomous software development. Poolside trains the compact model to continuous…
product-demo-director — Turn product evidence and raw footage into narration-paced, source-bound demo films—with Codex or Claude Code directing
product-demo-director is an open-source Python tool that automates creating product demo films. The utility leverages AI coding assistant…
mac-thermalright-ai-monitor — AI agent (Claude Code + Codex) & system monitor for the Thermalright 9.16 LCD — native macOS
Mac-thermalright-ai-monitor is a native macOS system monitor and AI agent for Thermalright LCD hardware. The Swift-based utility integrat…
openhub — Terminal discovery hub and package manager for AI coding tools, MCP servers, and agent skills. Built with Python & Textu
Openhub is a terminal-based discovery hub and package manager for AI developer tools. Built with Python and Textual, the CLI tool helps d…
creator-vibe — A Codex skill that keeps human outcomes and creative intent ahead of overengineering.
creator-vibe is an open-source agent skill designed to keep human creative intent ahead of overengineering. The project provides structur…
HonorBox
HonorBox is an open-source storefront for selling digital products using GitHub and Stripe. The project hosts static storefronts on GitHu…
Megaphone
Megaphone is an open-source, on-device Mac dictation app powered by Apple silicon. The MIT-licensed utility uses Apple's SpeechAnalyzer a…
AgentLoop
AgentLoop is an open-source, zero-dependency Node.js framework for running isolated, multi-cycle AI coding agents. It spins up a fresh Co…
HOL Guard
HOL Guard is an open-source security firewall designed to protect systems from rogue AI agents. It intercepts agent actions in real time …
Transcript-of-Liang-Wenfengs-DeepSeek-Founder-4-Hour-Investor-Meeting — 梁文锋四小时投资人会议实录，因原文在微信上无法查看，特传一份github
A leaked transcript captures DeepSeek founder Liang Wenfeng discussing AGI and LLM strategy. The four-hour investor meeting transcript su…
moonsconfig — Open travel OS with Maya AI for calls, support chat, RFQs, vendor outreach, itinerary curation, route maps, packages, ho
Moonsconfig is an open-source travel operating system powered by an AI assistant called Maya. It handles customer support chats, vendor o…
agent-notify — Self-cleaning notification attention queue for parallel Claude Code / AI-agent terminal sessions on macOS
agent-notify is a self-cleaning notification queue for managing parallel AI agent terminal sessions on macOS. The open-source Python tool…
skill-view — Local-only web GUI for inspecting agent skills (SKILL.md) across user, project, plugin, cache, and marketplace sources
skill-view is a local-first web GUI for inspecting and managing agent skills. The TypeScript tool aggregates SKILL.md files across user, …
anamra — A native macOS capture-to-outcome workspace: screenshot, OCR, voice, AI-assisted drafts, local semantic memory, and conf
anamra is a native macOS productivity workspace featuring local semantic memory and AI-assisted drafting. The open-source Swift applicati…
awesome-mcp-open
awesome-mcp-open is a curated GitHub repository for Model Context Protocol tools and resources. The growing directory helps developers di…
Finn-loop — The Finn-loop: a 3-skill AI software factory for Claude Code — spec, build, review. Humans merge.
Finn-loop is an agentic software factory framework built for Claude Code. The open-source project structures development into a three-ste…
Cisco bets its small open cybersecurity models can outperform GPT-5.5 at vulnerability detection for a fraction of the cost
Cisco releases two open-source cybersecurity AI models designed for efficient vulnerability detection. The models reportedly detect rough…
dgx-spark-laguna — Serve Poolside Laguna S 2.1 (NVFP4) on the NVIDIA DGX Spark (GB10) without hanging your box. Working stack, crash-safe c
dgx-spark-laguna is a crash-safe configuration stack for running Poolside Laguna S 2.1 on NVIDIA DGX Spark hardware. It provides working …
NVIDIA Open Sources First GPU-Accelerated Medical Physics Simulation Framework
NVFlare-MedPhysics is a new open-source GPU-accelerated framework for medical physics simulation. NVIDIA releases the toolkit to help dev…
enterprise-prd-toolkit — 把金融級 PRD 方法論工程化成四個 Claude Skills：找洞 → 寫規格 → 產測試。含交易所提現、自營交易挑戰賽的完整 PRD 範例。
enterprise-prd-toolkit is an open-source collection of Claude Skills for generating financial-grade product requirement documents. The to…
podcast-shorts-factory — Ten cooperating AI agents that turn long podcasts into short-form videos, automatically. Free and open source, runs on f
Podcast Shorts Factory is an open-source Python tool that uses ten cooperating AI agents to automate short-form video creation. The syste…
xinchao-dynamic-mind — 独立、可自托管的 AI 动态心智状态引擎：驱动力、念头池、疲惫、睡眠与意图。
xinchao-dynamic-mind is a self-hosted state engine that simulates drives, fatigue, and sleep for AI agents. The Node.js-based project int…
handoff-skill — Claude skill: turn your current conversation into a complete handoff document so any LLM can pick up exactly where you l
Handoff-skill is an open-source Claude skill that packages active conversations into comprehensive handoff documents. Developers can use …
skills — Free, MIT-licensed agent skills from StarSling. ci-speedup audits your GitHub Actions workflows and reports, from real r
StarSling releases skills, an MIT-licensed open-source collection of AI agent capabilities. The repository includes tools like ci-speedup…
Humalike x Hermes
Humalike x Hermes is an open-source social intelligence plugin for the Hermes Agent framework. A single command equips autonomous agents …
Redential
Redential is an open-source CLI tool that generates verifiable developer credentials from local git histories. The tool reads your local …
pwa-sense-bridge — Privacy-first mobile sensor events for LLM-powered PWAs.
pwa-sense-bridge is a privacy-first JavaScript library connecting mobile device sensors to LLM-powered Progressive Web Apps. It captures …
coreclaw-api-directory — An unofficial, categorized directory of 118 CoreClaw Worker APIs for web scraping, automation, lead generation, e-commer
CoreClaw-api-directory is a curated collection of unofficial worker APIs for developers. The repository catalogs 118 endpoints spanning w…
memory-forest — A verifiable local memory architecture for long-running AI agents
Memory-forest is a local-first memory architecture providing verifiable provenance for long-running AI agents. It organizes knowledge in …
TabNexus — Turn tab overload into a local-first workspace with intent-first AI organization and Agent collaboration through MCP.
TabNexus is a local-first Chrome extension that uses intent-driven AI to organize browser tabs. The TypeScript-based tool features Model …
opencode-configs — OpenConfig — best OpenCode + oh-my-openagent (OmO) + OpenRouter AI coding agent config. Sisyphus · Claude · GPT · DeepSe
OpenConfig is a curated configuration package for integrating OpenCode with Oh-My-OpenAgent and OpenRouter. The tool unifies multiple LLM…
ALLY-Skill — 一个人格 AI Skill 实验：别人垂直赛道，我垂直人格。
ALLY-Skill is an experimental open-source tool for building personality-driven AI agent skills. The project targets content creators and …
India’s AI hackathon seeks to build local, open-source tech - Scroll.in
India is hosting a national AI hackathon to develop localized, open-source technology. The initiative focuses on building foundational mo…
Latent-MoE — Implementation of LatentMoE,Toward Optimal Accuracy per FLOP and Parameter in Mixture of Experts (Elango et al., NVIDIA
Latent-MoE is a drop-in PyTorch layer implementing NVIDIA's efficient Mixture of Experts architecture. The dependency-light module optimi…
moonlit-myriad — 专为 AI 玩家设计的单文件卡牌肉鸽 🌙
Moonlit-myriad is a single-file Python roguelike card game built specifically for AI enthusiasts. The project combines deck-building mech…