AI Agent Orchestration and Skills
The largest cluster on today's trending list is tooling that helps developers build, configure, and audit autonomous agents.
- mvschwarz/openrig - Build persistent agent networks from Claude Code, Codex, and Pi with shared roles and context. (+642 today)
- earendil-works/pi - Unified LLM API, agent loop, TUI, and coding agent CLI in one toolkit. (+298 today)
- garrytan/gstack - Opinionated Claude Code setup with 23 tools covering CEO, designer, QA, and more roles. (+104 today)
- ComposioHQ/awesome-claude-skills - Curated list of Claude skills, resources, and workflow tools. (+319 today)
- alirezarezvani/claude-skills - 380 skills and 30 agents for Claude Code, Codex, Cursor, and eight other coding agents. (+138 today)
- Shubhamsaboo/awesome-llm-apps - 100 or more AI agents, agent skills, and RAG apps that are free and open source. (+136 today)
- humanlayer/skills - Agent skills from HumanLayer. (+66 today)
- google/skills - Agent skills for Google products and technologies. (+30 today)
- ifixai-ai/iFixAi - Independent auditing of AI agents, answering whether an agent is doing what it is supposed to do in under 120 seconds. (+1492 today)
Context Management and LLM Gateway Tools
- mksglu/context-mode - Context window optimization for AI coding agents, with 98 percent tool output reduction, session memory, and routing across 17 platforms via MCP. (+362 today)
- thedotmack/claude-mem - Persistent cross-session context for agents: captures, compresses, and re-injects relevant memory. (+99 today)
- diegosouzapw/OmniRoute - Free MIT AI gateway with one endpoint covering 359 providers, 1200 or more models, quota-aware fallback, and token compression. (+311 today)
- modelcontextprotocol/servers - Official Model Context Protocol server implementations. (+190 today)
- MakazhanAlpamys/Soup - Fine-tune LLMs from a single YAML file; layer streaming trains an 8B model on a 4 GB laptop GPU. (+152 today)
Inference Engines and GPU Tooling
- NVIDIA/OpenShell - Safe, private Rust runtime for autonomous AI agents. (+2456 today)
- magnitudedev/magnitude - Self-optimizing open source inference engine for agents that compiles and tunes kernels for your exact hardware, running models up to 2x faster than llama.cpp. (+538 today)
- sgl-project/sglang - High-performance serving framework for large language models and multimodal models. (+40 today)
- tile-ai/tilelang - Domain-specific language for high-performance GPU, CPU, and accelerator kernel development. (+163 today)
AI-Powered Media Generation
- debpalash/VoiceStudio - Fully local, open source ElevenLabs alternative covering voice cloning, video dubbing, transcription, and audiobook creation in 646 languages. (+1284 today)
- harry0703/MoneyPrinterTurbo - Generate HD short videos from a topic or keyword using an automated AI workflow. (+576 today)
- heygen-com/hyperframes - Write HTML, render video, built for agents. (+627 today)
- Friedrich-M/UniMate - SIGGRAPH Asia 2026 research: one unified model to animate diverse skeletons. (+217 today)
- VectifyAI/PageIndex - Document index for vectorless, reasoning-based RAG. (+477 today)
Developer Workflow and Infra
- cursor/plugins - Cursor plugin specification and official plugins. (+150 today)
- openai/codex - Lightweight coding agent that runs in your terminal. (+157 today)
- ag-ui-protocol/ag-ui - Agent-User Interaction Protocol for bringing agents into frontend applications. (+67 today)
- oblien/openship - Self-hosted deployment platform. (+111 today)
- t8y2/dbx - 25 MB cross-platform database client for 100 or more databases with built-in AI and MCP server. (+739 today)
- PostHog/posthog - Self-driving product platform with AI observability, analytics, session replay, and more. (+29 today)
- cloudflare/vinext - Vite plugin that reimplements the Next.js API surface and deploys anywhere. (+23 today)
- pablostanley/yoinks - Download any video from your terminal, no ads. (+361 today)
- nyakang/nyaterm - A modern remote terminal workspace. (+8 today)
- Pumpkin-MC/Pumpkin - Fast, efficient Minecraft server host written in Rust. (+75 today)
Today's list is an unusually concentrated signal: builders are treating agent infrastructure (context management, skill libraries, auditing, and multi-provider routing) as the new application layer, not a novelty. The parallel surge in local inference and media-generation tools suggests that running capable models on your own hardware is quickly becoming a baseline expectation rather than a stretch goal.