last30days-skill ranks trends by engagement, not editors

A GitHub skill called last30days ranks trending posts across 17 platforms by actual engagement instead of editorial position, per its repo. That's a different lens than most trend tools, which mirror whatever editors already pushed to the top. The repo backs the approach with real usage: 4,822 forks and 175 merged pull requests, 122 of them from 52 outside contributors, since it went up on January 23.

The skill queries Reddit with comments, X, YouTube with full transcripts, TikTok, Instagram Reels, Hacker News, Polymarket, GitHub, Bluesky, LinkedIn, StockTwits, Threads, arXiv, Techmeme, Digg and Xiaohongshu in parallel, then merges duplicate stories across platforms before ranking them. The ranking signal is engagement itself: upvotes, likes, views, market odds.

Four sources work with zero setup: Reddit, Hacker News, Polymarket and GitHub. The rest need configuration, the repo notes.

Project health backs up the fork count. The codebase carries more than 2,700 tests at 84% coverage across 1,159 commits, the repo shows.

The catch

Ranking by raw engagement instead of editorial judgment means the tool surfaces whatever's loudest, not what's most accurate. Upvotes, views and Polymarket odds are as gameable as any editor's homepage. Watch whether results end up dominated by those four zero-config sources, since the other thirteen need setup most people won't bother with.

The project's climb to 4,822 forks and 175 merged pull requests by July 31 is what's pulling attention to a skill built back in January.