GitHub Trending Daily Report — 2026-02-28

🔥 Today at a Glance — Top 3 Themes

📅 Weekly Context — Changes Compared to the Past 7 Days

• Explosive Growth in AI Agents: Continuing from last week, 7 or more of today's top 10 trending repos are LLM agent-related. The reach of the Claude ecosystem (Anthropic, ruvnet, etc.) is particularly notable.
• Shell/TypeScript Strength: Beyond Python, TypeScript (deer-flow, claude-flow, GitNexus, airi) and Shell (obra/superpowers, anthropics/claude-code) are emerging as agent framework languages.
• Multi-agent → Production-Ready Trend: The phrase production-ready is increasingly appearing in experimental frameworks. obra/superpowers (1,549 ★/today) even presents its own methodology alongside the framework.
• Growing Chinese Developer Community Participation: Repositories from China — bytedance/deer-flow, datawhalechina/hello-agents, Wei-Shaw/claude-relay-service — are entering the top rankings.

🔎 Top 10 Spotlight Repositories

1. obra/superpowers — Shell · ⭐64,637 (+1,549 today)

An agentic skills framework & software development methodology that works.

More than just a library — an agent skills framework that includes a development methodology. If you want to systematize an agentic automation process for your team, start with the methodology docs.

2. abhigyanpatwari/GitNexus — TypeScript · ⭐6,153 (+1,327 today)

Client-side knowledge graph creator + Graph RAG Agent. Runs entirely in browser.

Visualize GitHub repos as a knowledge graph without a server, straight in the browser, then explore code with a Graph RAG agent. Ready to apply immediately for understanding legacy codebases.

3. D4Vinci/Scrapling — Python · ⭐17,948 (+1,127 today)

Adaptive Web Scraping framework — single request to full-scale crawl.

An adaptive scraping framework that handles everything from a single request to large-scale crawls. Can be connected directly to the data collection stage of an AI data pipeline.

4. muratcankoylan/Agent-Skills-for-Context-Engineering — Python · ⭐12,329 (+836 today)

Collection of agent skills for context engineering, multi-agent architectures.

Context engineering is emerging as the new prompt engineering. Use this as a reference for multi-agent debugging and optimization.

5. bytedance/deer-flow — TypeScript · ⭐21,783 (+692 today)

Open-source SuperAgent: research, code, create — with sandboxes, memory, tools, subagents.

ByteDance's open-source super-agent. Integrates sandboxes, memory, and sub-agents to handle complex tasks spanning minutes to hours.

6. moonshine-ai/moonshine — C · ⭐5,783 (+587 today)

Fast and accurate ASR for edge devices.

ASR specialized for edge devices. If you need a lightweight STT for a local AI pipeline, this is the top candidate.

7. ruvnet/claude-flow — TypeScript · ⭐15,516 (+545 today)

Agent orchestration platform for Claude — enterprise-grade, RAG integration, Claude Code/Codex native.

A multi-agent swarm deployment platform for Claude. RAG integration and distributed intelligence are its key strengths.

8. anthropics/claude-code — Shell · ⭐71,115 (+515 today)

Agentic coding tool in terminal — understands codebase, executes tasks via natural language.

Still going strong. Cementing itself as the de facto standard for terminal-based AI coding assistants.

9. ruvnet/ruvector — Rust · ⭐1,890 (+411 today)

High Performance, Real-Time, Self-Learning Vector Graph Neural Network & Database in Rust.

A self-learning vector GNN DB implemented in Rust. Growing interest as a memory/search layer for AI agents.

10. ruvnet/wifi-densepose — Python · ⭐8,953 (+362 today)

WiFi-based dense human pose estimation through walls using commodity mesh routers.

A bold project that estimates human poses through walls using commodity WiFi routers. A production implementation of InvisPose.

📌 Bonus Spotlight

• steipete/CodexBar (Swift, +221): View OpenAI Codex·Claude Code usage stats in your menu bar — no login required. A DevEx tool trend to watch.
• alibaba/OpenSandbox (Python, +107): A multi-language SDK + Docker/K8s runtime sandbox platform for running AI agents.
• Wei-Shaw/claude-relay-service (JavaScript, +74): An open-source relay service that unifies Claude·OpenAI·Gemini subscriptions.

✅ Next Actions — 3 Things to Follow This Week

1. Read the obra/superpowers methodology — This will serve as a systematic reference point when introducing an agent-based development workflow to your team. Start with the methodology docs alongside the Shell script structure.
2. Run abhigyanpatwari/GitNexus locally — Since it runs directly in the browser, try dragging your current project repo in and experimenting with knowledge graph visualization — no installation needed.
3. Study muratcankoylan/Agent-Skills-for-Context-Engineering — Context engineering is emerging as a core AI development competency for the first half of 2026. Use this repo's skill collection as a basis to review your own agent system.

📊 Data Source: GitHub Trending (daily) — as of 2026-02-28 09:00 KST