GitHub repositories taking off in the last 90 days
Ranked by how many more stars each repo got in the last 90 days than its pace over the 90 days before would predict. Star counts come from GitHub. Last updated 21 Nov 2026, 13:20 UTC.
?
- Stars gained
- New stars in the window. "all" means the repo is younger than the window, so the number is every star it has.
- Today
- Stars since midnight US Pacific time, so a partial day.
- Above pace
- How many more stars the repo got than its pace in the window before would predict. +500 means 500 more than usual.
- Momentum
- A daily score: 8 points per star and 2 per fork since midnight UTC, reduced for repos that are this busy every day.
- Speed-up
- Stars per day in the window divided by stars per day in the window before. "New" means the repo is younger than the window. "Under 14 days" (twice the window; "Under 2 years" for 1 year) means it is older than the window but too young for a full window before it to compare with.
- Suspected
- Repositories with at least 50 stars are left out as likely bought stars when they have no detected programming language and got those stars within a day of being created; or got them within 14 days while having no watchers, at most 1 fork, no detected language or under 1 KB of code other than HTML and CSS, and either 5 KB or less of files or an owner that is a personal account with at most 1 follower. Repositories with no detected language are also left out when, at any age, they have 5 KB or less of files and got at least 90% of their stars on a single day that ended more than a day ago; or have 5 KB or less of files and belong to a personal account with no followers that was created at most an hour before the repository. Any repository with at least 50 stars is also left out when it belongs to a personal account with no followers and either is under 3 days old with no watchers and at least 5 forks, or is under 30 days old with at least half as many forks as stars and 128 KB or less of files while that account is under a year old; it stays out while that account has no followers. Any repository with at least 50 stars that is under 30 days old is also left out for good when at least 2 other repositories of the same name were created within an hour of it, all owned by personal accounts with no followers. Some repositories are also left out, or kept in, by hand after a review of the daily lists, whatever the rules above say. See them.
- Bars
- Stars per day (per week for 90 days and 1 year). Bright bars are the selected window. Heights use a square-root scale, so quieter days stay visible next to a spike.
| # | Repository | Stars | Today | 7d | 30d | 90d | Above pace ↑ | Momentum | Speed-up | Weekly stars |
|---|---|---|---|---|---|---|---|---|---|---|
| 47601 |
Automatic quad remeshing tool |
3,718 | +2 | +24 | +114 | +408 | -1,694 | 82 | 0.2x, slowing | |
| 47602 |
A personal AI agent with memory, personality, and autonomy. |
6,877 | -1 | +5 | +81 | +688 | -1,697 | 0 | 0.3x, slowing | |
| 47603 |
Surge-style whole-home gateway and control plane for macOS with IPv4/IPv6 support— mihomo TUN, dnsmasq-powered DHCP/DNS, per-device routing, and an agent-friendly validation workspace. |
2,728 | 0 | +37 | +178 | +496 | -1,708 | 0 | Under 180 days | |
| 47604 |
Keyboard-first local Markdown notes with Vim motions, diagrams, and MCP integration. |
2,596 | 0 | +8 | +47 | +308 | -1,711 | 0 | 0.2x, slowing | |
| 47605 |
The fastest browser for AI agents to run browser automation, built for sharing your logged-in browser state with your AI agents, like Codex or Claude Code, without disturbing you. Zero cost, zero config. |
23,936 | +20 | +1,031 | +5,065 | +11,026 | -1,711 | 184 | 0.9x, slowing | |
| 47606 |
The Memory Layer for AI Agents - Drop-in memory infrastructure for AI agents and apps. Context that persists. Built for production. |
69,298 | +5 | +303 | +1,698 | +5,718 | -1,716 | 59 | 0.8x, slowing | |
| 47607 |
Grok 账号注册流程自动化与管理工具,支持 WebUI/GUI/CLI、多邮箱服务、多线程并发、账号级多协议代理池、grok2api 入池、SSO 风控筛查与CPA OIDC 凭证导出。 |
2,430 | 0 | +4 | +89 | +341 | -1,716 | 8 | 0.2x, slowing | |
| 47608 |
SQLite extension + bindings for Postgres NOTIFY/LISTEN semantics with durable queues, streams, pub/sub, and scheduler |
3,053 | 0 | +4 | +15 | +56 | -1,719 | 0 | 0.0x, slowing | |
| 47609 |
谷歌新书Agent设计模式(agentic design patterns)最佳中文版,持续优化。附:在线阅读、pdf和epub电子书下载。 |
8,549 | -1 | +40 | +199 | +787 | -1,720 | 5 | 0.3x, slowing | |
| 47610 |
Your AI workspace for models, agents, and productivity. |
53,719 | +11 | +190 | +958 | +3,341 | -1,722 | 109 | 0.7x, slowing | |
| 47611 |
🧑🚀 Authentication and authorization infrastructure for SaaS and AI apps, built on OIDC and OAuth 2.1 with multi-tenancy, SSO, and RBAC. |
14,961 | +1 | +25 | +228 | +615 | -1,723 | 24 | 0.3x, slowing | |
| 47612 |
GPT-Image-2 PPT Generator Skill for Creating Image-Based PowerPoint Presentations in Codex and Other Skill-Compatible Agents |
8,081 | +5 | +179 | +874 | +2,967 | -1,727 | 80 | 0.6x, slowing | |
| 47613 |
Qlib is an AI-oriented Quant investment platform that aims to use AI tech to empower Quant Research, from exploring ideas to implementing productions. Qlib supports diverse ML modeling paradigms, including supervised learning, market dynamics modeling, and RL, and is now equipped with https://github.com/microsoft/RD-Agent to automate R&D process. |
50,470 | +6 | +180 | +902 | +2,798 | -1,728 | 116 | 0.6x, slowing | |
| 47614 |
A list of developer portfolios for your inspiration |
27,790 | +5 | +89 | +501 | +1,715 | -1,740 | 35 | 0.5x, slowing | |
| 47615 |
Let Claude (or any LLM) actually watch a video — scene-aware, deduplicated frames + transcript, from a URL or local file. Runs locally, MIT. |
2,329 | +1 | +11 | +74 | +282 | -1,743 | 8 | Under 180 days | |
| 47616 |
Convert PDF to markdown + JSON quickly with high accuracy |
41,009 | +3 | +85 | +452 | +2,052 | -1,744 | 52 | 0.5x, slowing | |
| 47617 |
Harness Engineering 学习指南 — 从概念理解到独立实践的深度学习档案 |
6,309 | +1 | +23 | +138 | +736 | -1,746 | 26 | 0.3x, slowing | |
| 47618 |
The batteries-included agent harness. |
31,495 | +6 | +204 | +1,015 | +3,357 | -1,748 | 128 | 0.7x, slowing | |
| 47619 |
A Python port of Pi’s minimalist coding agent. |
3,067 | 0 | +16 | +92 | +647 | -1,751 | 0 | Under 180 days | |
| 47620 |
Curated CS fundamentals for placement prep: DSA,Computer Networks, DBMS & SQL, OOPs, Operating Systems, System Design & Software Engineering |
3,469 | +1 | +18 | +103 | +846 | -1,751 | 8 | Under 180 days | |
| 47621 |
An open-source model family for long-form speech, dialogue synthesis, voice design, sound effects, and real-time streaming TTS |
4,355 | 0 | +27 | +122 | +364 | -1,756 | 0 | 0.2x, slowing | |
| 47622 |
Robust Speech Recognition via Large-Scale Weak Supervision |
113,203 | +14 | +459 | +2,201 | +5,814 | -1,758 | 118 | 0.8x, slowing | |
| 47623 |
🚀 World's largest GPT Image 2 prompt library, updated daily — 2000+ curated prompts with preview images, 16 languages. OpenAI's next-gen image model with pixel-perfect text rendering, cross-image consistency, and commercial-grade illustration. Free & open source. |
10,320 | +1 | +30 | +190 | +1,000 | -1,758 | 7 | 0.4x, slowing | |
| 47624 |
基于 AI Agent + MCP 工具链 + 渗透 Skill 编排, 配合大语言模型, 自然语言输入 → 自动完成「信息收集 → 漏洞发现 → 漏洞利用 → 报告生成」全流程。 |
3,950 | +1 | +26 | +183 | +1,038 | -1,765 | 7 | 0.4x, slowing | |
| 47625 |
ChatGPT 官网接口纯协议逆向实现,Go 语言编写,高并发、低内存、高性能。指纹级请求还原,支持 GPT-Image-2.6 生图/改图、GIF 动图、文本模型与语音合成/转写,兼容 OpenAI 接口协议,在线批量生图,号池管理,支持导入 CPA、sub2api 号池,可接入 Cherry Studio、New API 等软件 |
6,969 | +1 | +68 | +281 | +1,104 | -1,774 | 0 | 0.4x, slowing | |
| 47626 |
Code for Machine Learning for Trading, 3rd edition — from data sourcing to live execution. |
21,929 | +4 | +108 | +434 | +1,448 | -1,775 | 55 | 0.4x, slowing | |
| 47627 |
Terminal coding agent powered by Kimchi's multi-model orchestration |
2,290 | 0 | +4 | +22 | +117 | -1,775 | 0 | 0.1x, slowing | |
| 47628 |
slime is an LLM training framework for RL Scaling. |
8,851 | 0 | +27 | +144 | +666 | -1,778 | 7 | 0.3x, slowing | |
| 47629 |
Multi-protocol proxy management panel built with Python & FastAPI, deployable on Railway. |
4,362 | 0 | +14 | +91 | +1,276 | -1,780 | 0 | Under 180 days | |
| 47630 |
3,000+ reusable AI agent skills for Claude Code, Codex, Cursor and Gemini CLI. AAS Core adds local catalog discovery, agent-owned selection and plan preview. CLI, MCP and plugins. |
50,076 | +10 | +349 | +1,801 | +5,170 | -1,785 | 129 | 0.7x, slowing | |
| 47631 |
Open Source Computer Vision Library |
91,665 | +2 | +99 | +444 | +1,370 | -1,788 | 31 | 0.4x, slowing | |
| 47632 |
:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP. |
40,999 | +1 | +206 | +663 | +2,482 | -1,789 | 31 | 0.6x, slowing | |
| 47633 |
No description |
1,839 | 0 | -1 | +3 | +8 | -1,797 | 0 | 0.0x, slowing | |
| 47634 |
YAO = Yielding AI Outcomes. A rigorous engineering, evaluation, governance, and portability system for reusable agent skills. |
2,739 | 0 | +6 | +35 | +276 | -1,797 | 0 | 0.1x, slowing | |
| 47635 |
A collection of 100+ specialized Claude Code subagents covering a wide range of development use cases |
26,714 | +5 | +154 | +797 | +2,316 | -1,801 | 76 | 0.6x, slowing | |
| 47636 |
self-hosted, distributed Durable Objects |
6,352 | 0 | +32 | +289 | +2,252 | -1,809 | 0 | 0.6x, slowing | |
| 47637 |
Supports comfyui/API calls/modelscope calls |
3,196 | +1 | +10 | +51 | +346 | -1,819 | 24 | 0.2x, slowing | |
| 47638 |
Point it at any web page and it finds the video, transcodes it, and streams it to your TV in real time. It even burns in subtitles. |
2,546 | 0 | +5 | +65 | +346 | -1,823 | 0 | 0.2x, slowing | |
| 47639 |
Project NOMAD is an offline-first knowledge and education server. Wikipedia, thousands of books, courses, maps, and optional local AI, all running on hardware you own with no internet required. |
44,198 | +12 | +660 | +3,483 | +8,152 | -1,830 | 89 | 0.8x, slowing | |
| 47640 |
Open-source backend-as-a-service. Postgres, auth, storage, functions, AI gateway, MCP. |
3,033 | 0 | +2 | +9 | +586 | -1,838 | 0 | 0.2x, slowing | |
| 47641 |
LMCache: Supercharge Your LLM with the Fastest KV Cache Layer |
12,405 | +1 | +39 | +259 | +1,149 | -1,845 | 34 | 0.4x, slowing | |
| 47642 |
Run the full 2.78-trillion-parameter Kimi K3 model, DeepSeek V4.1 Flash or GLM-5.3-Flash beyond available RAM by streaming activated weights directly from NVMe. A dependency-free, embeddable C inference engine. |
2,575 | +1 | +16 | +71 | +352 | -1,850 | 24 | Under 180 days | |
| 47643 |
Beautiful map components. 100% Free, Zero config, one command setup. |
12,404 | 0 | +14 | +114 | +620 | -1,851 | 0 | 0.3x, slowing | |
| 47644 |
The ultimate RAG for your monorepo. Query, understand, and edit multi-language codebases with the power of AI and knowledge graphs |
5,392 | -1 | +37 | +117 | +630 | -1,852 | 0 | 0.3x, slowing | |
| 47645 |
Desktop app to generate 3D models from images or prompt using local AI — runs entirely on your GPU |
8,737 | +4 | +80 | +440 | +1,590 | -1,854 | 44 | 0.5x, slowing | |
| 47646 |
LLM inference server with continuous batching & SSD caching for Apple Silicon — managed from the macOS menu bar |
23,793 | +5 | +148 | +832 | +3,475 | -1,860 | 220 | 0.7x, slowing | |
| 47647 |
A trilingual (繁中 / English / 简中) learning roadmap for agentic AI: from LLM basics to multi-agent systems, with 240+ curated resources and hands-on examples. 中文 AI agent 學習地圖。 |
8,842 | 0 | +63 | +983 | +2,615 | -1,870 | 0 | 0.6x, slowing | |
| 47648 |
Skills for Design Engineers |
11,273 | +14 | +200 | +1,231 | +3,810 | -1,872 | 183 | 0.7x, slowing | |
| 47649 |
The open-source agent for CI/CD reliability. |
13,801 | +76 | +971 | +1,423 | +3,066 | -1,878 | 1,435 | 0.6x, slowing | |
| 47650 |
Focus on prompting and generating |
53,967 | +1 | +39 | +414 | +1,783 | -1,912 | 44 | 0.5x, slowing |
Today counts stars since midnight US Pacific time, so it is a partial day.