Denoise

This category is heavily dominated by tooling and frameworks designed to enhance Claude Code, focusing on contextual awareness, skill aggregation, and practical integration into developer workflows. It signals a strong commitment to making AI agents functional for coding.

• Fincept-Corporation/FinceptTerminal

  FinceptTerminal is a finance application offering market analytics and investment research. Its repeated top ranking signifies sustained interest in AI tools for data-driven financial decision-making, showcasing a strong vertical application of AI.

  repeated⭐ 12k
• forrestchang/andrej-karpathy-skills

  This repo provides a single `CLAUDE.md` file designed to improve Claude Code's behavior. It's noteworthy for addressing common LLM coding pitfalls through direct prompt engineering, highlighting a practical approach to agent refinement.

  repeated⭐ 74k
• zilliztech/claude-context

  Claude-context is a code search MCP for Claude Code, enabling the entire codebase as context for coding agents. This new tool addresses a fundamental challenge in large-scale AI coding, pushing the practical limits of LLM context windows.

  new⭐ 6.9k
• thedotmack/claude-mem

  Claude-mem is a Claude Code plugin that automatically captures, compresses, and injects relevant context from past coding sessions. It's significant for enhancing AI agent memory and continuity, a crucial step for more effective long-term programming assistance.

  new⭐ 65k
• SimoneAvogadro/android-reverse-engineering-skill

  This repo offers a Claude Code skill specifically for Android app reverse engineering. It exemplifies the trend of highly specialized AI agents tackling complex, domain-specific engineering tasks, reducing manual effort in tedious processes.

  new⭐ 4.5k
• VoltAgent/awesome-agent-skills

  This is a curated collection of over 1000 agent skills compatible with various AI coding platforms like Claude Code and Gemini CLI. Its daily trend highlights the rapid growth and community-driven expansion of AI agent capabilities and interoperability.

  new⭐ 17k
• multica-ai/multica

  Multica is an open-source platform for managed agents, allowing users to assign tasks, track progress, and compound skills for coding agents. It signals a move towards orchestrating AI agents as collaborative teammates in development workflows.

  new⭐ 19k
• coleam00/Archon

  Archon is an open-source harness builder for AI coding, aiming to make AI coding deterministic and repeatable. Its cooling trend suggests ongoing interest in robust, predictable AI development environments, addressing reliability concerns in agentic workflows.

  cooling⭐ 19k
• Donchitos/Claude-Code-Game-Studios

  This project transforms Claude Code into a game development studio with 49 AI agents and a hierarchical coordination system. Its cooling trend indicates that while ambitious, such complex agentic setups are still experimental and seeking broader adoption.

  cooling⭐ 15k
• affaan-m/everything-claude-code

  This repo focuses on optimizing agent harness performance for Claude Code and other platforms, covering skills, memory, and security. Its cooling trend suggests a foundational area that requires continuous refinement as AI coding platforms evolve.

  cooling⭐ 164k
• z-lab/dflash

  DFlash implements Block Diffusion for Flash Speculative Decoding. While its description is concise, its new weekly trend suggests it's a technical contribution in efficient LLM inference, addressing speed and resource optimization in model deployment.

  new⭐ 2.1k
• pascalorg/editor

  Editor enables creation and sharing of 3D architectural projects. Its presence, despite cooling, points to niche applications of graphical editors that might eventually integrate AI for design generation or collaboration, but is not directly an AI trend itself.

  cooling⭐ 14k

The category shows a blend of foundational agent frameworks, educational resources for building agents, and specialized applications like AI SRE. The repeated presence of `NousResearch/hermes-agent` along with new offerings from OpenAI and self-evolving agent concepts, indicates a continued push for more capable and autonomous AI systems.

• NousResearch/hermes-agent

  Hermes-agent, an agent designed to grow with the user, maintains its top rank. This repeated presence underscores ongoing research into adaptive and personalized AI agents, aiming for long-term user engagement and evolving capabilities.

  repeated⭐ 109k
• lsdefine/GenericAgent

  GenericAgent is a self-evolving agent that grows its skill tree, achieving full system control with significantly less token consumption. This new project represents a leap towards truly autonomous and efficient AI agents capable of learning on their own.

  new⭐ 5.7k
• EvoMap/evolver

  Evolver is a GEP-Powered Self-Evolution Engine for AI Agents, focusing on Genome Evolution Protocol. This new repo highlights the advanced research into meta-learning for agents, aiming for AI systems that can independently improve their own architectures and behaviors.

  new⭐ 6.4k
• microsoft/ai-agents-for-beginners

  This Microsoft project offers 12 lessons to get started building AI agents. Its daily trend indicates a strong demand for educational resources as the AI agent paradigm becomes more accessible, lowering the barrier to entry for developers.

  new⭐ 58k
• mvanhorn/last30days-skill

  This AI agent skill researches and synthesizes information from various platforms over 30 days. Its cooling trend suggests practical utility but also that the core concept of generalized web research agents is maturing, with new innovations needed to stand out.

  cooling⭐ 23k
• HKUDS/DeepTutor

  DeepTutor is an agent-native personalized learning assistant. Its cooling trend suggests that while personalized AI tutoring is a compelling application, widespread adoption of sophisticated agent-based educational systems is still in early stages.

  cooling⭐ 21k
• openai/openai-agents-python

  This is a lightweight, powerful framework from OpenAI for multi-agent workflows. Its new weekly trend signifies OpenAI's continued investment in enabling complex agentic architectures, providing foundational tools for developers in this space.

  new⭐ 24k
• Tracer-Cloud/opensre

  OpenSRE is an open-source toolkit for building AI SRE agents. This new project highlights the application of AI to operational tasks, aiming to automate Site Reliability Engineering through intelligent agents, an emerging use case for AI.

  new⭐ 2.2k

This diverse category features new applications of AI in niche areas, including privacy-preserving human sensing, financial market analysis, and advanced voice synthesis. The emergence of `omi` also points to the development of highly context-aware personal AI assistants that interact directly with a user's digital environment.

• siddharthvaddem/openscreen

  OpenScreen is an open-source alternative to Screen Studio for creating screen demos. Its cooling trend suggests a steady demand for free, high-quality productivity tools, even if not directly AI-driven, offering value in creative workflows.

  cooling⭐ 32k
• ruvnet/RuView

  RuView uses WiFi DensePose to turn WiFi signals into real-time human pose estimation and vital sign monitoring, without video. This novel approach emphasizes privacy-preserving sensing, offering a significant alternative to traditional camera-based systems.

  new⭐ 49k
• google-ai-edge/gallery

  This gallery showcases on-device ML/GenAI use cases, allowing local model testing. Its cooling trend suggests the concept of on-device AI remains important, but adoption might be constrained by specific hardware capabilities or developer tooling.

  cooling⭐ 22k
• shiyu-coder/Kronos

  Kronos is a foundation model for the language of financial markets. Its repeated trend confirms persistent interest in specialized large models tailored for complex domains like finance, indicating deep learning's continued penetration into specific industries.

  repeated⭐ 20k
• dayanch96/YTLite

  YTLite is a flexible enhancer for YouTube on iOS. This new utility project highlights continuous user demand for customized media consumption experiences, often through community-driven enhancements to popular platforms.

  new⭐ 4.9k
• Lordog/dive-into-llms

  《动手学大模型Dive into LLMs》 is a practical programming tutorial series for large language models. Its new weekly trend shows a strong educational demand for hands-on resources as developers seek to understand and implement LLM technologies.

  new⭐ 33k
• OpenBMB/VoxCPM

  VoxCPM2 is a tokenizer-free TTS model for multilingual speech generation, voice design, and cloning. Its repeated trend emphasizes the ongoing innovation in speech AI, particularly towards more flexible and high-fidelity voice synthesis without traditional tokenizers.

  repeated⭐ 15k
• sansan0/TrendRadar

  TrendRadar is an AI-driven public opinion and trend monitor with multi-platform aggregation and smart alerts. This new project addresses information overload, demonstrating the value of AI in filtering and synthesizing public data for actionable insights.

  new⭐ 54k
• jamiepine/voicebox

  Voicebox is an open-source voice synthesis studio. This new project signals increasing interest in accessible, customizable tools for audio creation, potentially leveraging AI for advanced features in a broader creative ecosystem.

  new⭐ 22k
• virattt/ai-hedge-fund

  AI-Hedge-Fund aims to build an AI hedge fund team. This project represents the aspirational goal of fully autonomous AI in complex financial operations, reflecting high expectations for AI's investment capabilities.

  new⭐ 57k
• microsoft/VibeVoice

  VibeVoice is Microsoft's open-source frontier voice AI. Its cooling trend suggests that while advanced voice AI is a strategic area, the market for general-purpose high-fidelity voice models might be stabilizing, or new features are needed for renewed excitement.

  cooling⭐ 41k
• BasedHardware/omi

  Omi is an AI that sees your screen, listens to conversations, and provides recommendations. This new project pushes the boundary of proactive, context-aware personal AI assistants, raising questions about privacy and integration into daily digital life.

  new⭐ 12k
• microsoft/markitdown

  Markitdown is a Python tool for converting files and office documents to Markdown. Its repeated trend indicates a consistent need for efficient document conversion, leveraging Python for practical automation in content management.

  repeated⭐ 114k
• onyx-dot-app/onyx

  Onyx is an open-source AI platform with advanced chat features supporting various LLMs. Its cooling trend suggests that while open-source AI chat platforms are valuable, market differentiation often relies on specific integration or unique capabilities beyond core chat.

  cooling⭐ 28k
• google-ai-edge/LiteRT-LM

  LiteRT-LM from Google AI Edge focuses on lightweight, real-time LLMs. Its cooling trend indicates continued work on efficient on-device AI, but also that progress in this specialized area is often iterative rather than explosively trending.

  cooling⭐ 4.1k
• hacksider/Deep-Live-Cam

  Deep-Live-Cam offers real-time face swap and one-click video deepfake. Its cooling trend suggests that while deepfake technology remains interesting, the ethical implications and potential for misuse often lead to cautious adoption or specialized applications.

  cooling⭐ 92k

Although small, this category highlights a growing emphasis on user control over AI models and data, alongside specialized data acquisition for niche markets like Polymarket. It reflects concerns about vendor lock-in and the need for tailored data strategies.

• thunderbird/thunderbolt

  Thunderbolt promotes user control over AI models and data to eliminate vendor lock-in. This project reflects a significant concern about data sovereignty and model choice, pushing for open, flexible AI ecosystems.

  new⭐ 3.6k
• warproxxx/poly_data

  Poly_data is a Polymarket data retriever that fetches, processes, and structures market data. This new tool showcases the specific need for specialized data acquisition and preparation for AI models in niche, real-time prediction markets.

  new⭐ 1.5k

This category focuses on critical infrastructure for RAG systems and long-horizon agent harnesses. `RAG-Anything` and `bytedance/deer-flow` underscore the ongoing challenge of providing robust, extended context and memory to AI, crucial for complex, multi-step tasks.

• HKUDS/RAG-Anything

  RAG-Anything is an all-in-one RAG framework. This new project addresses the fragmentation in RAG implementations, signaling a push towards more comprehensive and versatile solutions for knowledge retrieval and grounding in AI systems.

  new⭐ 17k
• bytedance/deer-flow

  Deer-flow is an open-source long-horizon SuperAgent harness for research, coding, and creation, utilizing sandboxes, memories, and subagents. Its cooling trend highlights the complexity and ongoing development required for truly autonomous, multi-tasking AI agents.

  cooling⭐ 63k