GitHub Trending Weekly Digest — March 2-7, 2026

Welcome to this week's GitHub Trending digest! We've analyzed 6 days of trending repositories (March 2-7), deduplicated them, and ranked by persistence. This week showcased 21 remarkable projects spanning AI agents, security tools, infrastructure, and specialized development frameworks.

🔥 Persistent Dominators (4+ Days on Trending)

AIRI — Your Self-Hosted Digital Companion

🔗 github.com/moeru-ai/airi
Trending: 4 days (Mar 2, 3, 5, 6) | Stars: 736+

What it does: AIRI is a self-hosted AI virtual companion platform inspired by Neuro-sama. It goes far beyond text-based chatbots, offering real-time voice conversations, gameplay integration (Minecraft, Factorio), and true digital companionship.

Why it matters: While platforms like Character.ai limit interactions to text, AIRI provides a complete digital life container. Your AI companion can chat, play games with you, watch your screen, and be genuinely present 24/7. It's the "cyber pet" experience you've been waiting for — fully open-source and self-hosted, giving you complete control over your digital companion's personality and capabilities.

Tech: TypeScript, Vue.js, WebGPU, WebAudio, WebAssembly, Tauri, Live2D/VRM, ElevenLabs TTS, DuckDB WASM, Mineflayer, xsAI, Model Context Protocol (MCP)

---

🚀 Multi-Day Contenders (2-3 Days)

Agency Agents — Your AI Dream Team

🔗 github.com/msitarzewski/agency-agents
Trending: 3 days (Mar 4, 5, 7) | Stars: 503+

What it does: A complete AI agent team suite featuring 55+ specialized personas — from frontend wizards to Reddit ninjas, from creative ideators to reality checkers. Each agent comes with professional expertise, unique personality, clear deliverables, and proven workflows.

Why it matters: Startups and small teams often lack domain experts. Agency Agents provides a ready-to-deploy AI expert team covering all bases — engineering, design, marketing, product, project management, testing, support, and spatial computing. It dramatically lowers project launch costs, giving small teams the power of a "full-stack agency."

Tech: Claude Code integration, Markdown configs, multi-agent orchestration, knowledge management, task automation

---

WiFi DensePose — See Through Walls

🔗 github.com/ruvnet/wifi-densepose
Trending: 2 days (Mar 2, 4)

What it does: Revolutionary human pose estimation using ordinary WiFi signals — no cameras, no wearables. Analyzes WiFi Channel State Information (CSI) to reconstruct body positions, breathing rates, and heart rhythms.

Why it matters: Privacy-friendly human sensing that works through walls and in complete darkness. Perfect for healthcare monitoring, elderly care, disaster rescue, and smart homes. It's the future of non-intrusive presence detection.

Tech: Rust (810x performance boost), Python, WebAssembly, ESP32-S3, Docker, Kubernetes, DensePose UV mapping, RuVector AI framework, signal processing (SpotFi, Hampel filter, Fresnel zone modeling)

---

OpenSandbox — Safe Execution for AI Agents

🔗 github.com/alibaba/OpenSandbox
Trending: 2 days (Mar 2, 3) | Stars: 1,179+

What it does: Alibaba's open-source universal sandbox platform providing multi-language SDKs, unified sandbox APIs, and Docker/Kubernetes runtimes for AI applications.

Why it matters: AI agents need secure, isolated environments to run code, manipulate files, and execute commands. OpenSandbox standardizes sandbox interfaces for coding agents, GUI agents, agent evaluation, and reinforcement learning training — with distributed scheduling at scale.

Tech: Python, Java/Kotlin, TypeScript/JavaScript, C#/.NET, Docker, Kubernetes, FastAPI, Code Interpreter, VNC, Playwright, VS Code (code-server)

---

Trivy — Comprehensive Security Scanner

🔗 github.com/aquasecurity/trivy
Trending: 2 days (Mar 3, 5) | Stars: 145+

What it does: All-in-one security scanner for containers, Kubernetes, code repos, and cloud environments — finding vulnerabilities, misconfigurations, and secrets.

Why it matters: Modern teams need multi-layered security checks: OS vulnerabilities (CVEs), dependency issues, infrastructure misconfigurations (IaC), leaked secrets (API keys, passwords), and license compliance. Trivy delivers unified scanning across container images, filesystems, Git repos, VM images, and Kubernetes clusters — all dimensions, one scan.

Tech: Go, SBOM, CVE databases, IaC scanning engine

---

Shannon — Autonomous AI Security Hunter

🔗 github.com/KeygraphHQ/shannon
Trending: 2 days (Mar 4, 5) | Stars: 484+

What it does: Fully autonomous AI hacker finding real vulnerabilities in web applications. Shannon achieves 96.15% success rate on the XBOW benchmark with source-aware, zero-prompt testing.

Why it matters: Traditional security testing is manual, time-consuming, and error-prone. Shannon autonomously discovers vulnerabilities without human prompting, analyzing source code directly. Its 96.15% success rate proves real-world utility — helping teams find and fix security holes before launch.

Tech: TypeScript, AI security analysis engine, static source code analysis, vulnerability detection algorithms, web security testing framework

---

💡 Single-Day Spotlights

Anthropic Prompt Engineering Interactive Tutorial

🔗 github.com/anthropics/prompt-eng-interactive-tutorial
Trending: Mar 2 | Stars: 60+

Official Anthropic interactive tutorial covering Claude prompt engineering from basics to advanced — 9 comprehensive chapters with hands-on exercises and example playgrounds.

---

Ruflo — Claude Agent Orchestration Platform

🔗 github.com/ruvnet/ruflo
Trending: Mar 2 | Stars: 766+

Leading Claude agent orchestration platform supporting multi-agent swarms, autonomous workflow coordination, and conversational AI systems — with distributed swarm intelligence and native Claude Code/Codex integration.

---

Superset — Multi-Agent Development IDE

🔗 github.com/superset-sh/superset
Trending: Mar 3 | Stars: 637+

The IDE for the AI agent era — run 10+ coding agents in parallel using git worktrees for isolated branches and working directories. Built-in diff viewer and task monitoring dashboard for seamless multi-agent development.

---

LMCache — Fastest KV Cache for LLMs

🔗 github.com/LMCache/LMCache
Trending: Mar 3 | Stars: 140+

The fastest KV cache layer for large language models, dramatically reducing TTFT (time to first token) and boosting throughput. Store reusable KV caches across GPUs, CPUs, disks, even S3 — achieving 3-10x latency savings with vLLM integration.

---

worldmonitor — Real-Time Global Intelligence Dashboard

🔗 github.com/koala73/worldmonitor
Trending: Mar 4 | Stars: 860+

AI-powered real-time global intelligence dashboard aggregating news, geopolitical monitoring, and infrastructure tracking in one unified situational awareness interface.

---

everything-claude-code — Battle-Tested Claude Code Configs

🔗 github.com/affaan-m/everything-claude-code
Trending: Mar 4 | Stars: 451+

Complete Claude Code configuration collection — agents, skills, hooks, commands, rules, MCPs. From an Anthropic hackathon winner's real-world setup. Skip the configuration headache and use battle-tested best practices.

---

SEO Machine — AI-Powered SEO Content Workspace

🔗 github.com/TheCraigHewitt/seomachine
Trending: Mar 5

Professional SEO content creation workspace built on Claude Code, integrating research, writing, optimization, and analysis workflows. Generates 2000-3000+ word SEO-optimized blog posts with Google Analytics 4 and Search Console integration.

---

Qwen-Agent — Enterprise-Grade LLM Agent Framework

🔗 github.com/QwenLM/Qwen-Agent
Trending: Mar 6

Agent framework and application suite built on Qwen 3.0+, supporting Function Calling, MCP, Code Interpreter, RAG, and Chrome extensions. Powers Qwen Chat with optimized 1M token context and superior RAG performance.

---

HVE Core — Microsoft's Hypervelocity Engineering Framework

🔗 github.com/microsoft/hve-core
Trending: Mar 6

Microsoft's open-source Hypervelocity Engineering core framework providing enterprise-grade prompt engineering libraries for GitHub Copilot — 34 professional agents, 68 coding instructions, 40 reusable prompt templates, validated artifacts with JSON Schema.

---

CyberStrikeAI — AI-Native Security Testing Platform

🔗 github.com/Ed1s0nZ/CyberStrikeAI
Trending: Mar 6

AI-native security testing platform built in Go, integrating 100+ security tools with intelligent orchestration, role-based testing, skill systems, and full lifecycle management. Native MCP protocol support with 12+ security roles.

---

AReaL — Lightning-Fast RL Training for Agents

🔗 github.com/inclusionAI/AReaL
Trending: Mar 6

Lightning-fast reinforcement learning training system for large reasoning models and agents, jointly developed by Tsinghua IIIS and Ant Group. Fully open-source with 2.77× training acceleration and SOTA performance across math, coding, search, and customer service domains.

---

MiroFish — Swarm Intelligence Prediction Engine

🔗 github.com/666ghj/MiroFish
Trending: Mar 7

Universal swarm intelligence engine predicting anything. Next-gen AI prediction engine using multi-agent technology to build high-fidelity parallel digital worlds from real-world seed information — for policy rehearsal, sentiment forecasting, narrative prediction, and financial analysis.

---

OpenAI Skills — Codex Agent Skills Catalog

🔗 github.com/openai/skills
Trending: Mar 7

OpenAI Codex's skills catalog. Agent Skills are folders containing instructions, scripts, and resources that AI agents can discover and use for specific tasks — "write once, use anywhere."

---

Google Cloud Generative AI Examples

🔗 github.com/GoogleCloudPlatform/generative-ai
Trending: Mar 7

Google Cloud generative AI sample code and notebook collection using Gemini on Vertex AI — complete learning path covering Gemini 3.1 Pro, Vertex AI Search, RAG, Imagen, Chirp, and more.

---

Jido — Elixir Autonomous Agent Framework

🔗 github.com/agentjido/jido
Trending: Mar 7

Autonomous agent framework for Elixir, built for distributed systems, autonomous behavior, and dynamic workflows. Based on pure functional agent design and OTP runtime with formalized agent patterns.

---

📊 Weekly Themes

This week's trending repositories revealed three dominant themes:

AI Agents Everywhere

• Orchestration platforms: Ruflo, Agency Agents, OpenSandbox
• Framework maturity: Qwen-Agent, Jido, HVE Core
• Specialized agents: SEO Machine, CyberStrikeAI, Shannon
• Agent infrastructure: OpenAI Skills, Superset, LMCache

Security Automation Rising

• AI-powered testing: Shannon (96.15% vulnerability detection)
• Comprehensive scanning: Trivy (containers, K8s, IaC, secrets)
• Native AI integration: CyberStrikeAI (100+ tools, MCP protocol)
• Shift-left security: Automated, continuous testing at scale

Privacy-First Innovation

• WiFi DensePose: Camera-free pose estimation
• Self-hosted companions: AIRI (your data, your control)
• Open-source infrastructure: Alibaba OpenSandbox, Tsinghua AReaL
• Developer sovereignty: Everything runs on your infrastructure

---

🎯 Takeaway

This week marks a turning point in AI development: agents are becoming specialized, standardized, and production-ready. We're moving from generic chatbots to professional AI teammates with clear roles, proven workflows, and measurable outcomes.

The rise of agent orchestration platforms (Ruflo, Agency Agents), standardized skill systems (OpenAI Skills), and enterprise-grade frameworks (Qwen-Agent, HVE Core) signals that AI agents are no longer experimental — they're shipping to production.

Security is getting smarter too. Tools like Shannon and CyberStrikeAI demonstrate that AI can autonomously find real vulnerabilities with human-level (or better) accuracy, shifting security testing from annual audits to continuous, automated validation.

Most importantly, the open-source community is embracing privacy-first, self-hosted solutions. Projects like AIRI and OpenSandbox prove you don't need to sacrifice privacy for cutting-edge AI capabilities.

The future is autonomous, specialized, and open. And it's trending now.

---

Compiled by Tommy Zhang | March 8, 2026