GitHub Trending Weekly Digest — June 9-13, 2026

Welcome to this week's GitHub Trending roundup, covering June 9-13, 2026. This digest consolidates the most persistent trending repositories, ranked by how many days they stayed on the trending list. We've deduplicated entries and translated them into polished, readable English for your convenience.

🏆 Trending Champions (4+ Days)

addyosmani/agent-skills — Production Engineering for AI Coders

Days trending: 4 (June 10-13)
🔗 github.com/addyosmani/agent-skills

What it does: A comprehensive skill pack containing 24 structured workflows that enforce production-grade engineering practices for AI coding agents. It includes 7 slash commands (/spec, /plan, /build, /test, /review, /code-simplify, /ship) that map to the complete development lifecycle.

Why it matters: AI agents naturally take shortcuts — skipping specs, tests, security reviews, and other best practices. This project codes senior engineer discipline into structured workflows with anti-rationalization tables (listing common AI "excuses" and rebuttals), red flag detection, and evidence-based verification requirements. Based on Google's "Software Engineering at Google" practices.

Tech: Markdown skill files, Claude Code plugin system, supports Cursor / Gemini CLI / Windsurf / OpenCode / Codex / GitHub Copilot

🔥 Multi-Day Dominators (2-3 Days)

apple/container — Native Mac Containers

Days trending: 3 (June 11-13)
🔗 github.com/apple/container

What it does: Apple's official lightweight container tool for macOS, written in Swift. Creates and runs Linux containers through lightweight VMs, fully compatible with OCI standards.

Why it matters: The first official Apple-backed container runtime optimized for Apple Silicon. Provides native container experience without Docker Desktop overhead, leveraging macOS 26's latest virtualization and networking enhancements.

Tech: Swift, macOS 26 Virtualization framework, Containerization Swift package, OCI image spec

tolaria — Files-First Knowledge Management

Days trending: 3 (June 9-10, 12)
🔗 github.com/refactoringhq/tolaria

What it does: Cross-platform desktop app (macOS/Windows/Linux) for managing Markdown knowledge bases. Files-first, Git-first, offline-first, and AI-first — designed for second brains, team docs, and AI context storage.

Why it matters: Zero vendor lock-in. Uses pure Markdown + YAML frontmatter with Git versioning. No proprietary formats, no cloud dependency, no forced structure. The developer personally manages 10,000+ notes with it daily. Natively integrates with Claude Code, Codex CLI, and Gemini CLI.

Tech: Tauri 2, React, TypeScript, Rust, WebKit2GTK (Linux), Git, pure Markdown

last30days-skill — Cross-Platform Research Agent

Days trending: 2 (June 9-10)
🔗 github.com/mvanhorn/last30days-skill

What it does: AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, TikTok, Instagram, and GitHub. Uses real engagement metrics (upvotes, bets, comments) as signal strength instead of SEO rankings. Generates comprehensive reports with full citations.

Why it matters: Traditional search engines aggregate editorial content but miss real community discussions scattered across walled gardens. Google can't search Reddit comments or X posts; ChatGPT has Reddit but not X/TikTok; Gemini has YouTube but not Reddit; Claude has none. This skill searches all platforms at once, merges duplicate stories, and synthesizes findings with entity-aware pre-research (v3 feature resolves people/orgs/products before searching).

Tech: Python 3.12+, yt-dlp, Node.js (Bird client for X), ScrapeCreators API, SQLite (trend monitoring with --store), Claude Code / Codex / Cursor / Gemini CLI support

maziyarpanahi/openmed — Local Medical AI

Days trending: 2 (June 11-12)
🔗 github.com/maziyarpanahi/openmed

What it does: Open-source medical AI toolkit with 1,000+ models for disease detection, drug extraction, and PII de-identification — running 100% locally. Supports 12 languages and 247 PII detection checkpoints.

Why it matters: Medical data privacy is critical. Cloud APIs require uploading patient data to third-party servers. OpenMed keeps everything on-device, HIPAA-compliant, with MLX acceleration (24-33x faster on Apple Silicon). Includes native iOS/macOS support via OpenMedKit Swift library.

Tech: Python, PyTorch, Apple MLX (Apple Silicon acceleration), Swift (OpenMedKit), Hugging Face transformers

NVIDIA/SkillSpector — AI Skill Security Scanner

Days trending: 2 (June 11, 13)
🔗 github.com/NVIDIA/SkillSpector

What it does: Security scanner for AI agent skills, detecting vulnerabilities and malicious patterns before installation. Two-stage analysis: fast static scan (regex + AST + YARA) + optional LLM semantic evaluation.

Why it matters: Research shows 26.1% of AI agent skills contain vulnerabilities, 5.2% show malicious intent. SkillSpector detects 64 vulnerability patterns across 16 categories (prompt injection, data exfiltration, privilege escalation, supply chain attacks, etc.) and provides 0-100 risk scores with clear recommendations (SAFE / CAUTION / DO NOT INSTALL).

Tech: Python, LangGraph (workflow), YARA (signature matching), AST analysis, OSV.dev API (real-time CVE lookup), optional LLM semantic evaluation

phuryn/pm-skills — Product Manager AI Toolkit

Days trending: 2 (June 10-11)
🔗 github.com/phuryn/pm-skills

What it does: Product Manager skills marketplace with 68 PM skills and 42 chained workflows across 9 plugins. Encodes proven PM frameworks (Teresa Torres' Opportunity Solution Trees, Marty Cagan's INSPIRED, Alberto Savoia's lean prototyping) as structured AI workflows.

Why it matters: Generic AI only generates text. PM Skills provides structure — step-by-step guidance through discovery, hypothesis validation, prioritization, and execution. Commands can chain together (e.g., /discover suggests /write-prd next). Includes OKRs, RICE/ICE prioritization, Lean Canvas, user journey mapping, and more.

Tech: Claude Code & Cowork native plugins, Markdown skill files, also supports Codex / Cursor / Gemini CLI / Kiro

⚡ Single-Day Standouts

turbovec — 16x Vector Compression

Day: June 9
🔗 github.com/RyanCodrai/turbovec

What it does: Rust vector indexing library based on Google's TurboQuant algorithm. Compresses 10M documents from 31GB to 4GB (16x), searches faster than FAISS, with zero training required.

Why it matters: RAG systems suffer from memory explosion and complex training. TurboQuant eliminates training overhead while delivering better recall. Hand-written SIMD kernels (ARM NEON / x86 AVX-512) outperform FAISS FastScan by 12-20% on ARM. Search-time allowlist filtering solves multi-tenant RAG challenges with zero recall loss.

Tech: Rust, SIMD (ARM NEON / x86 AVX-512), Python bindings (Maturin), TurboQuant (random rotation + Lloyd-Max quantization + length renormalization)

supervision — Computer Vision Swiss Army Knife

Day: June 9
🔗 github.com/roboflow/supervision

What it does: Reusable detection annotation, tracking, and dataset conversion tools for computer vision projects. Model-agnostic post-processing layer with 20+ customizable annotators.

Why it matters: Model inference is just the first step. The real work is visualization, tracking, dataset conversion, and region analysis — supervision eliminates this repetitive wheel-reinventing.

Tech: Python 3.9+, compatible with Ultralytics, Transformers, MMDetection, ByteTrack, FTS5 full-text search

opencv — The Computer Vision Standard

Day: June 9
🔗 github.com/opencv/opencv

What it does: Industry-standard open-source computer vision library with 2,000+ optimized algorithms for image processing, object detection, face recognition, OCR, and machine learning.

Why it matters: 30+ years of maturity, used by Google, Microsoft, Intel, Apple. The foundational infrastructure for CV, cited in tens of thousands of academic papers. Commercial-friendly BSD license.

Tech: C++ core, Python/Java/MATLAB bindings, CUDA/OpenCL/Intel IPP/ARM NEON acceleration, built-in DNN module

soxoj/maigret — Username Intelligence

Day: June 10
🔗 github.com/soxoj/maigret

What it does: OSINT tool that finds accounts across 3,000+ websites using only a username. Auto-extracts profile info, recursively searches linked accounts, and generates multi-format reports (HTML/PDF/Xmind/JSON/CSV/interactive graphs). Optional AI analysis mode generates neutral investigation summaries (real name / location / occupation / interests / confidence / follow-up leads).

Why it matters: Essential for security researchers, investigators, and OSINT practitioners mapping digital footprints across platforms.

Tech: Python 3.10+, async network requests, optional FlareSolverr (Cloudflare bypass), optional OpenAI-compatible API (AI analysis), Web UI (Flask)

masterking32/MasterDnsVPN — Censorship-Breaking DNS Tunnel

Day: June 12
🔗 github.com/masterking32/MasterDnsVPN

What it does: Advanced DNS tunneling VPN with ultra-low overhead (5-7 bytes vs 59B DNSTT / 24B SlipStream). Automatic retransmission, multi-path support, and intelligent resolver selection.

Why it matters: Battle-tested during Iran's 88-day internet lockdown — when traditional VPNs and physical connectivity failed, MasterDnsVPN stayed connected. 9x faster than DNSTT, 3.6x faster than SlipStream.

Tech: Go, custom protocol + ARQ, AES/ChaCha20/XOR encryption, Docker, supports Linux/Windows/macOS/Android/embedded (ARM/MIPS/RISC-V)

kenn-io/agentsview — AI Agent Session Analytics

Day: June 13
🔗 github.com/kenn-io/agentsview

What it does: Local-first session analytics for 20+ AI coding tools (Claude Code, Codex, Cursor, Copilot CLI, Gemini CLI, etc.). Indexes sessions into SQLite for 100x faster querying than re-parsing raw files.

Why it matters: Tracking usage and costs across multiple AI tools is difficult. Existing tools (like ccusage) re-parse files every time. agentsview indexes once, searches instantly with FTS5, and provides token usage/cost dashboards.

Tech: Go, SQLite (FTS5), Svelte 5, TypeScript, optional PostgreSQL/DuckDB backends

iptv-org/iptv — Public IPTV Channels

Day: June 13
🔗 github.com/iptv-org/iptv

What it does: Collection of publicly available IPTV channels from around the world. Aggregated M3U playlists with EPG (Electronic Program Guide) support.

Why it matters: Centralizes user-submitted public live stream URLs for use in any video player that supports streaming.

Tech: TypeScript, M3U playlists, EPG (via iptv-org/epg), channel database (iptv-org/database)

📊 This Week's Themes

AI Tooling Maturity
The explosion of AI agent skills ecosystems (agent-skills, pm-skills, last30days-skill) signals a shift from raw model conversations to modular, composable skill packs. The emergence of SkillSpector reflects growing security concerns — 26% vulnerability rate demands trust infrastructure.

Local-First Renaissance
Three projects (tolaria, openmed, MasterDnsVPN) emphasize data sovereignty and zero cloud dependency. Whether it's knowledge management, medical privacy, or censorship resistance, users are reclaiming control over their data.

Apple's Developer Ecosystem Push
Apple officially entering the container space (apple/container) and MLX acceleration powering medical AI (openmed) shows Apple Silicon becoming a serious developer platform beyond consumer apps.

Performance Optimization Deep Dive
Projects like turbovec (hand-written SIMD kernels for 16x compression) and agentsview (100x query speedup) demonstrate that extreme optimization still matters — even in the AI era.

🎯 Takeaway

This week showcased a maturing AI tooling ecosystem where quality, security, and privacy are no longer afterthoughts. The most persistent trending projects aren't flashy prototypes — they're production-grade infrastructure solving real pain points: engineering discipline (agent-skills), data sovereignty (tolaria), platform optimization (apple/container), and trust verification (SkillSpector).

Compiled by Tommy Zhang | June 14, 2026