Skip to content

What’s trending on GitHub today

The open-source projects winning the most stars right now. Open any of them to see what it does and how to install it.

Collected

The full list

Each row is one project. The green number is how many stars it gained today; the number in the first column is its position on GitHub.

16 repositories

  1. Position 1

    elementalsouls/Claude-BugHunter

    A Claude Code skill bundle for bug hunting and external red-team work - 82 skills, 15 slash commands, 681 disclosed-report patterns curated across 24 core vulnerability classes, plus enterprise identity + infrastructure attack matrices.

    Python14.7k stars in total1.7k forks#LLM

    +1,622

    stars today

    +12%

  2. Position 2

    obra/superpowers

    An agentic skills framework & software development methodology that works.

    Shell323k stars in total28.7k forks#AI agents

    +890

    stars today

    +0.3%

  3. Position 3

    HKUDS/CLI-Anything

    "CLI-Anything: Making ALL Software Agent-Native" -- CLI-Hub: https://clianything.cc/

    Python61.6k stars in total5.6k forks

    +773

    stars today

    +1.3%

  4. Position 4

    jingyaogong/minimind

    🧠 两小时从 0 训出你的 64M 大模型。Train your own 64M LLM from scratch in 2 hours.

    Python67.8k stars in total8.9k forks#LLM

    +847

    stars today

    +1.3%

  5. Position 5

    cursor/plugins

    Cursor plugin specification and official plugins

    TypeScript17.9k stars in total2k forks

    +299

    stars today

    +1.7%

  6. Position 6

    firecrawl/firecrawl

    Knowledge infrastructure to make your agents smarter. Search, scrape and access 550+ data providers through one API.

    TypeScript206k stars in total10.7k forks#AI agents#LLM

    +1,241

    stars today

    +0.6%

  7. Position 7

    nginx/nginx

    The official NGINX Open Source repository.

    C32.6k stars in total8.4k forks

    +246

    stars today

    +0.8%

  8. Position 8

    openvinotoolkit/openvino

    OpenVINO™ is an open source toolkit for optimizing and deploying AI inference

    C++11.8k stars in total3.6k forks#LLM

    +217

    stars today

    +1.9%

  9. Position 9

    cathrynlavery/diagram-design

    Editorial diagram design for Claude Code, Codex, GitHub Copilot, Cursor, Factory Droid, Qoder, and Pi. 45 diagram types. Self-contained HTML + SVG. No shadows. No Mermaid slop.

    HTML69.4k stars in total4.7k forks#AI agents

    +636

    stars today

    +0.9%

  10. Position 10

    mvanhorn/last30days-skill

    AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary

    Python71.6k stars in total6.3k forks#AI agents#LLM

    +681

    stars today

    +1.0%

  11. Position 11

    google-labs-code/stitch-skills

    A library of Agent Skills designed to work with the Stitch MCP server. Each skill follows the Agent Skills open standard, for compatibility with coding agents such as Antigravity, Gemini CLI, Claude Code, Cursor.

    TypeScript11.1k stars in total1.4k forks#MCP#AI agents

    +315

    stars today

    +2.9%

  12. Position 12

    mjmirza/app-store-compliance

    Avoid common App Store and Google Play rejections. Open playbook and pre-submission guard with 117 documented rejection patterns and fixes, privacy manifest and Data safety checks, 95 tracked deadlines (EU AI Act, DSA, COPPA, age laws), and a skill for AI coding agents.

    Shell2.9k stars in total183 forks#AI agents

    +869

    stars today

    +43%

  13. Position 13

    Tracer-Cloud/opensre

    The open-source agent for CI/CD reliability.

    Python13.7k stars in total1.9k forks#LLM

    +466

    stars today

    +3.5%

  14. Position 14

    toeverything/AFFiNE

    There can be more than Notion and Miro. AFFiNE(pronounced [ə‘fain]) is a next-gen knowledge base that brings planning, sorting and creating all together. Privacy first, open-source, customizable and ready to use.

    TypeScript75.9k stars in total5.6k forks

    +676

    stars today

    +0.9%

  15. Position 15

    XiaoYouChR/Ghost-Downloader-3

    The only downloader you need. 下载器的集大成者。

    Python13.9k stars in total849 forks

    +1,082

    stars today

    +8.4%

  16. Position 16

    zhouxiaoka/autoclip

    AutoClip|一个链接,一键出片。开源 AI 视频剪辑桌面工具,将播客、访谈、课程等长视频自动剪成短视频,生成字幕、封面和发布文案,适配抖音、小红书、TikTok、Reels 与 YouTube Shorts。Open-source AI video clipping & content repurposing.

    Python12.4k stars in total2k forks#MCP#AI agents

    +979

    stars today

    +8.6%

Order, star totals and “stars gained” are the ones GitHub published, read at Nov 21, 2026, 06:58 UTC. Nothing in this list is estimated or re-ranked by us.

Source: github.com/trending · updated

How to read a row

Three numbers tell the story of each project. These are the ones of elementalsouls/Claude-BugHunter, first on the list.

Position
The place GitHub gave the project on its trending page. We copy the order; we do not rank anything ourselves.
Stars gained today
A star is how people on GitHub bookmark a project they like. This is how many new ones it received in the period.
Growth
The gain compared with the stars the project already had. Small projects on the rise show the largest percentages.

Worked calculation: daily star velocity & growth

See how the figures on this page are computed, step by step, using real numbers from today’s trending list.

Open in comparison tool →
Select project to calculate:

Step 1 · Base stars before today

13,080 stars

14,702 (total)- 1,622 (gained today)= 13,080 baseline count

GitHub only reports total stars and stars gained. Subtracting the gain reveals the starting point at the beginning of the period.

Step 2 · Relative growth rate

+12%

1,622 (gain) / 13,080 (base)= +12%
12% proportion

An established repository with a large baseline shows smaller percentage changes, whereas an emerging project shows higher relative growth from the same daily gain.

Step 3 · Average hourly velocity

~67.6 stars / hour

1,622 stars / 24 hours= ~67.6 stars per hour

The steady pace needed to accumulate 1,622 stars over 24 hours. A surge from a viral post often exceeds this rate during peak hours.

Comparing #1 elementalsouls/Claude-BugHunter vs #2 obra/superpowers

Two projects from today’s list demonstrating contrasting momentum profiles.

Gained today
+1,622
Growth
+12%
Hourly pace
~67.6/h
Gained today
+890
Growth
+0.3%
Hourly pace
~37.1/h

Compare license, maintenance freshness, open issues and recorded star history side by side:

Compare Claude-BugHunter vs superpowers →

How to read this list

Each line is one repository, in the position GitHub gave it. The number in the first column is that position. The green figure on the right is the number of stars GitHub says the repository gained today, and the percentage under it compares that gain with the stars the repository already had. A project that jumps from 300 to 900 stars and one that goes from 90,000 to 90,600 both gained 600; the percentage is what tells them apart.

The title opens a page for the repository with its description, the first paragraph of its README, license, dates and, when the README has one, the command to install it. Details shows the same facts without leaving the list.

What “trending” does and does not tell you

A place on this list means many people starred the repository in a short time. It says nothing about whether the code is maintained, documented or safe to run. Launch-day attention, a popular post or a well-known author can put a project here before it has a single release. Before depending on something you found here, look at the last push date, the open issues and the license.

GitHub does not explain how it orders the page, and the order is not simply “most new stars first”: you can check that in the list above by sorting it by Stars gained and comparing with GitHub order.

Side-by-side comparison of trending alternatives

When multiple projects in the same domain trend together, you can compare them side by side in our comparison tool. It matches their stars gained today, relative growth rate, hourly velocity, SPDX license, open issues count, and recorded star history in an aligned matrix.

Other ways to look at the same data

The weekly and monthly lists smooth out one-day spikes. Fastest growing reorders today’s repositories by relative growth, which favours small projects on the rise. The archive by date keeps the list of each past day, and How it works explains where every number comes from.

Why use this instead of the GitHub page

Every list says when it was read
The collection time is printed above each list, and a warning appears when it is more than 24 hours old. A list is never shown as “today” without its date.
The ranking is GitHub’s, not ours
Positions and “stars today” are copied from github.com/trending. There is no hidden score; the method page describes each number.
A page for every day
The daily list is stored under its date in the archive, so yesterday’s ranking does not disappear when today’s replaces it.
Install command only when one exists
The command box appears only when the README contains a real install line that names the project. No generic “git clone” on every card.
Filters you can link to
Search text, language, topic, sort order and view are written to the address, so a link reproduces exactly what you see.
Feeds for readers and scripts
The daily list, languages and topics have their own RSS and JSON feeds, with the same collection time on every item.

More lists to explore

The same data cut by period, programming language and subject.

Questions and answers

More detail on the method page.

Where does this list come from?

From the public page github.com/trending. A scheduled job on our server reads that page, keeps the repositories in the order GitHub shows them, and stores the result with the time of the reading. That time is printed above the list.

How often is it updated?

The collector is scheduled to run every three hours. If a run fails, or GitHub returns a page we cannot read, the previous list stays on screen with its own collection time, and a warning appears once that time is more than 24 hours old.

How does GitHub decide what is trending?

GitHub does not publish the formula. The page shows how many stars each repository gained in the period, but the order does not follow that number strictly: a repository with fewer new stars is often placed above one with more. We copy the order as it is and do not try to reproduce it.

Why are there fewer than 25 repositories?

GitHub lists at most 25 repositories per page and often shows fewer. This site shows exactly the repositories that were on the page when it was read, without padding the list from other sources.

What does the percentage under “stars today” mean?

It is the relative growth: stars gained in the period divided by the stars the repository had before the period. A project that went from 1,000 to 1,200 stars shows +20%. It is left blank when the repository had no stars before the period.

How does the worked example on this page calculate star velocity and growth?

The worked calculation subtracts stars gained today from total stars to establish the baseline count, divides stars gained by that baseline to determine relative growth rate, and divides today’s gain by 24 hours to estimate average hourly velocity. This highlights whether a project is experiencing a viral surge or steady accumulation.

How do I compare projects with very different total stars?

An established repository with a large baseline shows a small percentage increase, whereas a new repository gaining the same number of stars shows much higher relative growth. Use the Compare tool to evaluate absolute velocity, percentage growth, license, and maintenance side by side.

Where do license, issue counts and dates come from?

Metadata including SPDX license identifiers, open issue counts, creation dates and last push timestamps are read directly from GitHub’s public API and repository READMEs during collection runs, preserved without modification.

Can I share a filtered list?

Yes. Search text, language, topic, sort order and view are written to the address of the page as you change them. Copy the address and the same filtered list opens for anyone.