GitHub · Open Source

Top 20 GitHub Projects by 24-Hour Star Growth: AI Tooling, Reverse Engineering, and More

By miofly · Nov 5, 2026

Read original on juejin.cn ↗ Google Translate ↗ Alt translation

The 24-hour Star gains cluster around AI agent orchestration, security audits, and UI design quality. Open source effort is shifting from AI code generation to controlling and refining the output, and these projects set the standards that will shape AI-assisted development workflows.

Summary

GitHub's daily trending list for 2026-11-06 tracks 20 projects by their 24-hour Star increments. The top gainer is REA, a TypeScript reverse-engineering tool built on a unified MCP plugin framework, which added 3,585 Stars. Second place went to Craft Anything, a pure Rust engine that analyzes the implementation details of any website, app, or game and uses AI to generate a compilable clone. NVIDIA's auto-ontology framework, a Python tool for building semantic graphs from metadata, took third with +2,240.

The list clusters around AI-assisted development. mattpocock/skills offers composable engineering skills for Claude Code and Codex, covering requirement clarification, TDD, code review, and architecture improvement. Cloudflare's security-audit-skill defines a six-phase machine-readable security audit process. t3code is a control panel for managing multiple local AI coding agents across mobile, web, and desktop. Design-quality tools also stand out: impeccable enforces 23 commands and 59 deterministic rules against generic AI UI patterns, while unslop detects and fixes UI 'slop' in web interfaces. Other notable entries include Vercel Labs' agent-desktop for reliable desktop automation via OS accessibility trees, Scrapling for web scraping, Lichii a C++ Switch emulator, deepseek-harness a plugin-based agent toolkit, and omniget a download manager for 1,800+ sites.

Takeaways

— REA, a TypeScript reverse-engineering tool with an MCP plugin framework, led the daily list with +3,585 Stars.

— Craft Anything, a pure Rust engine, gained +3,082 Stars by using AI to analyze and clone websites and apps.

— NVIDIA's auto-ontology Python framework for semantic graph building added +2,240 Stars.

— mattpocock/skills provides composable engineering practice skills for Claude Code and Codex, gaining +2,092 Stars.

— Cloudflare's security-audit-skill defines a six-phase machine-readable audit process, up +2,059 Stars.

— t3code acts as a control panel for managing multiple local AI coding agents, up +1,961 Stars.

— Vercel Labs' agent-desktop uses OS accessibility trees for reliable desktop automation, up +1,793 Stars.

— Impeccable enforces design quality with 23 commands and 59 deterministic rules, up +1,216 Stars.

Conclusions

The dominance of AI coding skills and audit tools indicates a community move from trusting AI output to actively verifying and controlling it.

Both impeccable and unslop target generic AI-generated UI, showing that design quality is a recognized pain point with multiple solutions emerging.

Reverse engineering tools like REA and Craft Anything are gaining traction alongside AI agents, suggesting developers want to reuse and understand existing software rather than build from scratch.

Concepts & terms

slop

In AI-generated UI, slop refers to generic design patterns such as purple gradients, glowing effects, emoji tags, bold text, and uneven spacing that make interfaces look machine-made.

ADE

Agent Desktop Environment, a desktop environment for orchestrating multiple AI coding agents, often using Git worktrees to isolate tasks and provide a unified interface.

Source: juejin.cn ↗ Google Translate ↗ Backup ↗