Something worth noting across today's GitHub trending: the most-starred repos are increasingly collections of skills and tools you add to existing workflows rather than standalone applications you run in isolation. The pattern points toward AI capabilities becoming infrastructure — modular, reusable, and shared — rather than features locked inside a single product.
mvanhorn/last30days-skill: Real Engagement Over SEO
last30days-skill is an agent skill that researches any topic by pulling signals from Reddit, X, YouTube, Hacker News, Polymarket, and the broader web simultaneously, ranking results by genuine engagement — upvotes, view counts, prediction-market odds — rather than search-engine position. It picked up 2,535 stars today and sits near 40,000 total, reflecting real demand from developers building research agents that need to surface what people actually care about rather than what's been optimized to rank.
github.com/mvanhorn/last30days-skill
addyosmani/agent-skills: Production Engineering Capabilities for Any Coding Agent
agent-skills packages proven software-engineering capabilities — code review, refactoring, test generation, debugging — as modular, drop-in skills for AI coding agents. With 53,279 stars and 821 new additions today, the project reflects a shift in teams that have moved past proof-of-concept experiments and need reliable, composable primitives they can trust in production pipelines rather than bespoke prompt chains they maintain themselves.
github.com/addyosmani/agent-skills
harry0703/MoneyPrinterTurbo: AI Short-Form Video Generation at Scale
MoneyPrinterTurbo turns a text prompt or script into a polished short-form video, handling narration, b-roll selection, subtitles, and audio mixing through an AI pipeline. It gained 1,389 stars today and has now crossed 85,000 total — one of the most consistently trending repos in the AI video space. As the tooling around AI video matures, the gap between a content idea and a publishable clip keeps narrowing.
github.com/harry0703/MoneyPrinterTurbo
roboflow/supervision: A Standard Library for Computer Vision
supervision provides reusable utilities for building computer vision applications — annotation overlays, zone detection, tracking integrations, and dataset management — so developers stop rewriting the same infrastructure for every new project. The library added 695 stars today on its way past 43,000, a sign that computer vision is maturing in the same direction as the rest of the AI tooling ecosystem: toward composable, well-maintained abstractions rather than custom one-off implementations.
github.com/roboflow/supervision
LucaRonin/tolaria: A Desktop App for Markdown Knowledge Bases
tolaria is a native desktop application for building and navigating markdown-based knowledge bases — linking notes, managing structure, searching a local corpus — without sending any content to a cloud service. Its 612-star day brings it near 15,200 total, and the timing reflects renewed interest in local-first knowledge tools as teams grow more deliberate about where their information actually lives.