The 11 November edition ranks the settled 10 November window, and the top three keep yesterday’s order. morluto/rea stays first, the toolkit for pointing a coding agent at software you have no source for: it inspects native binaries through Hopper or Ghidra, JavaScript and Electron apps, .NET assemblies and websites, then explains how a feature works with the evidence and the limits behind each conclusion. meta/astryx holds second, the design system Meta grew internally over eight years and opened in beta with more than 150 React components, themes you restyle through CSS variables and a CLI that people and coding agents drive the same way. huamidev/GestaltX remains third, the MobileGestalt editor for iOS 27.x whose SwiftUI interface exposes hidden device switches, from the region code people flip to try enabling Apple Intelligence to Dynamic Island layout and always-on display, with an automatic backup before every write.
The middle of the chart reshuffles. janestreet/agent_skills climbs to four, Jane Street’s first public skill collection, currently one skill that turns your agent into a Socratic tutor with a diagnostic, a level map and spaced review. pingdotgg/t3code rises to five as a single control surface for the coding agents already on your machine, driving Claude Code, Codex, Cursor and others through your own subscriptions from a phone, a browser or a desktop app. sherlock-project/sherlock checks a username across more than 400 social networks at six, Minireel/HGhub still collects stars at seven even though its author removed all code, releases and build history days ago, storytold/craft-anything studies how any app or game works and then has an agent rebuild it at eight, Ebony-Vinyl/dsh-our-free-model puts free frontier models one install away inside the DeepSeek Harness shell at nine, and NotRexed/Hookshot closes the chart by recompiling the Ocarina of Time Switch 2 remake into a native PC program.
If you only try one project today, let janestreet/agent_skills tutor you for twenty minutes: the diagnostic alone shows where your knowledge map is thin, and the spaced review schedule is the part most self-study setups skip.