💎 Awesome GitHub Gems

A hand-curated collection of the most useful open-source repositories — for hackers, builders, and self-taught developers.

Cybersecurity · OSINT · AI · Self-hosting · Learning — sorted by what they actually do.

---

🧭 What is this?

A curated list of GitHub repositories that are genuinely worth your time — no filler. Each one is sorted by what it's for, with a one-line description so you know if it's useful before you click.

⚠️ Security note (read this): Some entries are AI "skills" / plugins that run code with your permissions when installed. Before installing any of them: review the code, distrust names that imitate official ones, and distrust suspiciously high star counts. Trusting = executing. These are marked with ⚠️.

---

📚 Contents

• 🔐 Cybersecurity & Pentesting
• 🕵️ OSINT & Web Scraping
• 🤖 AI Agents & Coding Assistants
• 🧠 AI & Machine Learning
• 🎬 Content Creation
• 🏠 Self-Hosted & Privacy
• 🧩 Code Understanding
• 📖 Learning & References
• 🎮 Game Development
• 🛡️ Offline & Survival
• 📦 Data Structure Libraries

---

🔐 Cybersecurity & Pentesting

• PayloadsAllTheThings — The legendary reference of web pentest payloads & bypasses (SQLi, XSS, SSRF, XXE...), with methodology for each. The web hacker's bible.
• BurpSuite-collections — 80+ third-party Burp Suite extensions, pentest PDFs and tips. ⚠️ Third-party extensions — review before loading.
• Anthropic-Cybersecurity-Skills — Hundreds of cybersecurity workflows for AI agents, mapped to MITRE ATT&CK / NIST. ⚠️ Not official Anthropic — audit before installing.

🕵️ OSINT & Web Scraping

• Scrapling — Adaptive Python scraping framework with anti-detection: stealth fetcher, Cloudflare bypass, fingerprint spoofing, proxy rotation. Great for OSINT pipelines.

🤖 AI Agents & Coding Assistants

⚠️ Many of these install into AI coding tools and run code with your permissions. Audit first.

• anthropics/skills — Official Anthropic skills for Claude. The trusted baseline.
• multica — Turn coding agents into real teammates: assign tasks on a board, agents execute autonomously. Supports Claude Code, Copilot, Pi & more.
• andrej-karpathy-skills — A CLAUDE.md of principles (inspired by Karpathy) that make AI write better code: think before coding, simplicity, surgical changes, verifiable goals. Safe (text only).
• superpowers — Composable software-development methodology for AI coding agents (spec refinement, planning, TDD, code review). ⚠️ Runs hooks.
• ECC — Agent optimization system for Claude Code: specialized agents, reusable skills, memory & context management. ⚠️ Runs hooks.
• taste-skill — Skills that upgrade AI-built UIs (better layout, typography, motion). Framework-agnostic.
• academic-research-skills — Claude Code skills for the full academic research pipeline: literature search, writing, peer review.
• pi — Agent harness: interactive coding agent, tool-calling runtime, and a unified multi-provider LLM API.
• hermes-agent — Autonomous AI agent that creates its own skills, keeps persistent memory, and runs across CLI/Telegram/Discord/WhatsApp.
• 500-AI-Agents-Projects — 500+ AI agent projects with code (LangGraph, CrewAI, AutoGen), organized by industry.

🧠 AI & Machine Learning

• ML-For-Beginners — Microsoft's 12-week, 26-lesson classic ML course (scikit-learn). No deep learning, no GPU required — perfect to start.
• whisper — OpenAI's open-source speech-to-text. Multilingual, runs small models on CPU.
• DeepTutor — Agent-native learning workspace: tutoring, quizzes, RAG knowledge bases, interactive books.

🎬 Content Creation

• MoneyPrinterTurbo — Generate short HD videos with AI from a single topic: script, subtitles, voice & music. Posts to TikTok/IG/YouTube Shorts.
• yt-dlp — The best video/audio downloader (YouTube + 1000s of sites).

🏠 Self-Hosted & Privacy

• vaultwarden — Lightweight Bitwarden server in Rust. Run your own encrypted password manager on anything (even a Raspberry Pi).
• bitwarden/server — The official Bitwarden server (heavier; vaultwarden is lighter).
• n8n — Self-hosted workflow automation (low-code). Like Zapier, but yours.
• plausible — Privacy-friendly web analytics (a Google Analytics alternative).
• cal.com — Open-source scheduling (a self-hosted Calendly).
• AppFlowy — Open-source Notion alternative.
• penpot — Open-source design & prototyping (a Figma alternative).

🧩 Code Understanding

• codegraph — Builds a local code index (SQLite + tree-sitter) so AI agents understand a codebase fast. 20+ languages.
• Understand-Anything — Turns any codebase into an interactive knowledge graph with guided learning tours.

📖 Learning & References

• awesome-python — The definitive curated list of the best Python libraries & resources.
• awesome-nodejs — Curated list of the best Node.js resources.
• system-design-primer — Learn how to design large-scale systems (famous for interviews & real building).
• the-art-of-command-line — Master the terminal — years of experience condensed.
• engineering-blogs — Huge list of engineering blogs from top companies (Netflix, Uber, Stripe...).
• how-they-load — How 60+ companies do load testing and scale their systems.
• backend-cheats — Visual cheat sheet of core backend topics (networking, OS, DBs, DevOps).
• RefactoringGuruExample — Design pattern code examples (from Refactoring Guru).
• angular-realworld-example-app — Well-structured "RealWorld" reference app to learn Angular properly.

🎮 Game Development

• open-source-games — Massive list of open-source games (with source) across ~18 categories. Learn game dev by reading real code (Godot, Bevy).

🛡️ Offline & Survival

• project-nomad — A 100% offline "survival" knowledge server: semantic AI, offline Wikipedia (Kiwix), education, maps, encryption. Docker, Apache 2.0.

📦 Data Structure Libraries

Useful only if you code in that language — they're libraries, not resource lists.

• doctrine/collections (PHP) · montagejs/collections (JavaScript) · eclipse-collections (Java) · apple/swift-collections (Swift) · Java-Projects-Collections (Java practice projects) · ordinals-collections (Bitcoin Ordinals/NFTs)

---

🔥 Into cybersecurity?

If you're into cybersecurity, check out my profile — and try OBSIDIAN, a free local OSINT framework I built.

It runs 100% on your machine: domain/IP/email/username recon, relationship graphs, timelines, mobile UI — no cloud, no tracking.

⭐ Star this list if it helped you · 🤝 PRs welcome — share your own gems