AI and LLM Tooling
- anthropics/claude-code: Agentic coding assistant living in your terminal, handling tasks and git workflows through natural language (+483 today)
- Fission-AI/OpenSpec: Spec-driven development framework designed for AI coding assistants (+367 today)
- cactus-compute/needle: Automation foundation model for tiny devices, 2-bit and 8-29 MB, running tool calls on phones, wearables, and microcontrollers (+234 today)
- NVIDIA/TensorRT-LLM: Python API for defining and optimizing LLM inference on NVIDIA GPUs (+14 today)
- unslothai/unsloth: Local UI for running and fine-tuning LLMs and diffusion models including GGUF and DeepSeek formats (+68 today)
- EricLBuehler/mistral.rs: Fast, flexible LLM inference engine written in Rust (+3 today)
- shiyu-coder/Kronos: Foundation model for the language of financial markets (+64 today)
AI Agent Memory and Context
- supermemoryai/supermemory: Fast, scalable memory and context engine that runs fully locally, positioned as the memory API for the AI era (+392 today)
- mem0ai/mem0: Drop-in memory infrastructure layer for AI agents and apps, built for production (+62 today)
- anthropics/knowledge-work-plugins: Open-source plugin library for knowledge workers using Claude Cowork (+281 today)
Browser Automation and Agent Environments
- Tencent/BrowserSkill: Lets AI agents use your real, logged-in browser via CLI and extension without interrupting your work (+612 today)
- browser-use/browser-use: Python library enabling agents to browse the web (+194 today)
- ChromeDevTools/chrome-devtools-mcp: Chrome DevTools exposed as an MCP server for coding agents (+39 today)
- yyjeqhc/webcodex: Gives cloud AI agents a real development environment on your own machines (+137 today)
Workflow Automation and Agent Orchestration
- n8n-io/n8n: Fair-code workflow automation platform with native AI capabilities and 400-plus integrations (+126 today)
- TencentCloud/Octop: Self-hosted, multi-user, multi-agent AI assistant (+281 today)
- withastro/flue: Sandbox agent framework from the Astro team (+22 today)
- google-labs-code/stitch-skills: Library of agent skills following the open Agent Skills standard, compatible with Claude Code, Cursor, and Gemini CLI (+9 today)
Document Processing and RAG
- docling-project/docling: Prepares documents of all types for generative AI pipelines (+129 today)
- opendatalab/MinerU: Converts PDFs and Office files into LLM-ready Markdown and JSON (+59 today)
- VectifyAI/PageIndex: Document index enabling vectorless, reasoning-based RAG (+24 today)
Web Frameworks and Generative UI
- vercel-labs/json-render: Generative UI framework for rendering AI-driven interfaces from JSON (+585 today)
- remorses/gpuix: Node.js and React bindings for Zed's GPUI, enabling native apps without Electron (+24 today)
Infrastructure and Storage
- rustfs/rustfs: Open-source, S3-compatible high-performance object storage, migration-friendly with MinIO and Ceph (+295 today)
- cloudflare/quiche: Cloudflare's QUIC transport and HTTP/3 implementation in Rust (+31 today)
- rui314/mold: Modern linker rewritten in Rust for faster build times (+57 today)
- tokio-rs/tokio: Async runtime for Rust providing I/O, networking, and scheduling (+8 today)
- pola-rs/polars: Extremely fast DataFrame query engine written in Rust (+12 today)
Dev Workflow and Productivity Tools
- ahmedkhaleel2004/gitdiagram: Generates free, fast, interactive diagrams for any GitHub repository (+285 today)
- EpicGames/lore: Next-generation open-source version control system from Epic Games (+23 today)
- Automattic/harper: Offline, privacy-first grammar checker powered by Rust (+75 today)
- yynxxxxx/Codex-X: Cross-platform visual management tool for OpenAI Codex CLI with provider switching, session sync, and MCP management (+32 today)
- ZuodaoTech/everyone-can-use-english: English learning tool for non-native speakers (+48 today)
Finance and Market Platforms
- Open-Dev-Society/OpenStock: Open-source alternative to paid market platforms with real-time prices and company insights (+472 today)
- virattt/ai-hedge-fund: AI-powered hedge fund team simulation (+91 today)
The throughline across today's list is clear: builders are racing to solve the hardest unsolved problems in agent systems, specifically persistent memory, reliable browser control, and grounding agents in real development environments. The simultaneous surge in generative UI tooling and document-to-LLM pipelines suggests that the plumbing layer for AI-native applications is maturing fast, and teams are moving from prototypes to production-grade infrastructure.