Denoise

A single entry from Anthropic, a major LLM provider, suggests a structured collection of examples for using their models in financial services. Its presence alone is a signal of enterprise focus.

• anthropics/financial-services

  A collection of resources from Anthropic for using their models in financial services. Its presence on the trending list signals a strong push by the model provider to capture enterprise use cases in a high-value vertical.

  rising⭐ 17k

The application layer for AI agents shows two distinct patterns: wrappers providing free access to proprietary models (free-claude-code, 9router) and specialized agents targeting high-value domains like financial research (dexter).

• Alishahryar1/free-claude-code

  A tool providing free terminal and VSCode access to Claude Code. This type of repository consistently trends, highlighting the community's persistent demand for free access to powerful proprietary models and the ongoing friction of API costs.

  cooling⭐ 23k
• virattt/dexter

  An autonomous agent specifically for financial research. The specialization of agents for high-value verticals like finance is a significant trend, moving beyond general-purpose coding assistants to domain-specific analytical tools.

  new⭐ 25k
• rowboatlabs/rowboat

  An open-source AI coworker with memory. The framing as a 'coworker' and the emphasis on memory positions it as a more integrated, long-term assistant rather than a simple command-line tool, reflecting ambitions for deeper workflow integration.

  new⭐ 14k
• decolua/9router

  A tool that routes requests to various free AI model providers to ensure uptime and avoid rate limits, claiming significant token savings. It's a pragmatic solution to the cost and reliability issues of using multiple free-tier AI services.

  rising⭐ 6.2k

This category is standardizing around the concept of 'skills'—curated, high-quality prompts and instructions for AI agents. It reflects a community effort to create reusable, shareable agent capabilities, moving beyond simple prompt engineering.

• forrestchang/andrej-karpathy-skills

  A single Markdown file of prompts designed to improve Claude's coding behavior, based on observations by Andrej Karpathy. It's an example of the 'skill library' pattern, where expert knowledge is codified into reusable prompts for agents.

  cooling⭐ 122k
• mattpocock/skills

  A collection of prompts for Claude from well-known developer Matt Pocock. Like other 'skills' repos, this represents a trend of developers sharing their personal, curated prompt libraries to improve AI agent performance on specific tasks.

  repeated⭐ 68k
• addyosmani/agent-skills

  A collection of prompts from Google engineer Addy Osmani, aimed at improving AI coding agents. This is another data point in the trend of creating standardized, shareable 'skills' to make agent behavior more reliable and effective.

  repeated⭐ 37k
• ComposioHQ/awesome-codex-skills

  This is a curated list of 'Codex skills' for workflow automation. It's another example of the community coalescing around the 'skill' as a primitive for interacting with and programming AI agents, though the repo itself is a simple list.

  cooling⭐ 7.8k

The trend here is open-source, self-hostable alternatives to commercial AI content generation services. Projects offer unfiltered, unrestricted creation of video, music (a Suno alternative), and presentations, catering to users seeking more control and lower costs.

• AIDC-AI/Pixelle-Video

  An automated short video generation engine. This project taps into the demand for programmatic content creation, particularly for social media. Its repeated trending indicates sustained interest in AI-driven video automation.

  repeated⭐ 14k
• hugohe3/ppt-master

  An AI tool that generates editable PowerPoint (PPTX) files from documents. The key feature is its creation of native shapes and animations, not just images, making it more useful than many similar tools.

  cooling⭐ 14k
• Anil-matcha/Open-Generative-AI

  A self-hostable, unrestricted AI image and video generation studio. The 'unrestricted' and 'no content filters' positioning is its main draw, appealing to users who want to avoid the safety limitations of commercial platforms.

  cooling⭐ 12k
• fspecii/ace-step-ui

  A user interface for an open-source AI music generation model, positioned as a free, local alternative to Suno. This follows a common pattern of open-source projects building UIs on top of powerful but hard-to-use models.

  new⭐ 3.5k
• OpenBMB/VoxCPM

  A tokenizer-free text-to-speech (TTS) model for multilingual generation and voice cloning. The 'tokenizer-free' aspect is technically significant, potentially allowing for better handling of diverse languages and vocal nuances.

  cooling⭐ 18k
• microsoft/markitdown

  A Python tool from Microsoft for converting various document formats to Markdown. While a simple utility, its origin from a major company and its utility for documentation workflows likely contribute to its visibility.

  cooling⭐ 122k

This is a Cambrian explosion of agent infrastructure. Multiple competing projects are building orchestration platforms, agent operating systems, and full-stack backends. The space is crowded and vying to become the foundational layer for agentic applications.

• bytedance/UI-TARS-desktop

  An open-source multimodal AI agent stack from ByteDance. This is a significant entry into the agent infrastructure space by a major technology company, signaling serious investment in building platforms for agentic applications.

  new⭐ 31k
• ruvnet/ruflo

  An agent orchestration platform specifically for Claude. It aims to help developers build and deploy multi-agent systems and workflows, representing a move towards more complex, coordinated agent behaviors rather than single, monolithic agents.

  new⭐ 47k
• TauricResearch/TradingAgents

  A framework for building multi-agent systems for financial trading. Its appearance highlights the strong interest in applying complex agentic architectures to high-stakes, high-value domains like automated trading.

  new⭐ 72k
• NousResearch/hermes-agent

  An agent project from Nous Research, a well-known group in the open-source AI community. While the description is vague, its origin suggests an attempt at building a more capable open-source agent.

  cooling⭐ 140k
• multica-ai/multica

  An open-source platform for managing coding agents as team members. It focuses on task assignment and progress tracking, abstracting away the low-level agent orchestration. This points to a future of human-agent teams.

  cooling⭐ 26k
• 1jehuang/jcode

  A 'Coding Agent Harness', likely a framework or testing environment for developing and evaluating coding agents. Its existence signals the need for standardized tooling to benchmark and improve the performance of different agent implementations.

  new⭐ 5.3k
• cocoindex-io/cocoindex

  An 'incremental engine for long horizon agents'. This tackles a key challenge in agent development: maintaining state and making progress on complex, long-running tasks. It represents a specialized component in the emerging agent stack.

  new⭐ 9.3k
• lsdefine/GenericAgent

  A self-evolving agent that claims to grow a 'skill tree' from a small seed to achieve system control. This project embodies the ambitious goal of agent autonomy and self-improvement, a central theme in agent research.

  cooling⭐ 10k
• browserbase/skills

  An SDK for building Claude agents that can browse the web. This is another example of the 'skills' abstraction, providing a pre-packaged tool (web browsing) that can be easily integrated into an agent.

  new⭐ 2.9k
• openai/symphony

  From OpenAI, this tool frames project work as 'autonomous implementation runs'. It's a high-level orchestration system for managing coding agents, reflecting a strategic move towards supervising agent outputs rather than agent processes.

  new⭐ 23k
• InsForge/InsForge

  A full backend platform for coding agents, providing database, auth, compute, and more. This project aims to be the 'Firebase for agents', providing all the necessary backend infrastructure so developers can focus on agent logic.

  new⭐ 9.2k
• Q00/ouroboros

  Pitched as an 'Agent OS', this project represents the ambition to create a foundational operating system layer for agents to run on, moving from ad-hoc scripts to a more structured, managed environment.

  new⭐ 3.8k

This category highlights the growing need for specialized, persistent memory for AI agents. These projects are moving beyond the limitations of context windows to provide long-term storage and retrieval, a critical component for complex, multi-session tasks.

• rohitg00/agentmemory

  A persistent memory system specifically designed for AI coding agents. Its focus on this single, critical component highlights the maturation of the agent stack, where dedicated solutions are replacing generic approaches for core functions like memory.

  new⭐ 3.1k
• thedotmack/claude-mem

  A plugin that captures and compresses Claude coding sessions to provide relevant context in future interactions. This is a practical implementation of agent memory, aiming to solve the context window limitation problem at the application layer.

  cooling⭐ 74k

This category shows a blend of traditional open-source SaaS alternatives like docuseal and new, AI-centric educational content from groups like datawhalechina. The standout is Google's chrome-devtools-mcp, adapting core developer tooling for the new paradigm of agent development.

• datawhalechina/hello-agents

  An educational tutorial on building intelligent agents from scratch. Its popularity reflects a strong demand from developers to learn the principles and practices of the rapidly growing agent development field.

  new⭐ 45k
• docusealco/docuseal

  An open-source alternative to DocuSign. Its trendiness amidst a sea of AI projects shows the enduring appeal of open-source solutions to common business problems and a desire for self-hosted, controllable software.

  new⭐ 16k
• datawhalechina/easy-vibe

  A programming course for beginners. Like other educational content from Datawhale, its popularity is driven by the organization's large community in China and consistent output of learning materials.

  new⭐ 8.2k
• ChromeDevTools/chrome-devtools-mcp

  A version of Chrome DevTools intended for use by coding agents. This is a major signal from Google, indicating that core developer tools are being adapted for a future where AI agents are the primary users.

  new⭐ 39k
• masterking32/MasterDnsVPN

  A VPN that uses DNS tunneling to bypass censorship. This is a niche but persistent category of tool that often trends due to regional internet restrictions, reflecting an ongoing global demand for uncensored access.

  new⭐ 2.2k
• playcanvas/supersplat

  A 3D Gaussian Splatting editor. This technology is a relatively new and efficient way to render 3D scenes, and the presence of an open-source editor indicates growing developer interest and adoption in the graphics community.

  new⭐ 6.0k
• Lordog/dive-into-llms

  An educational, hands-on tutorial series for learning about large language models, structured similarly to the popular 'Dive into Deep Learning' book. Its popularity stems from the high demand for practical, code-first LLM learning resources.

  new⭐ 36k
• oracle-devrel/oracle-ai-developer-hub

  A hub of technical resources from Oracle for building AI applications on its cloud and database services. This is a standard content marketing play from a large enterprise vendor to attract developers to their ecosystem.

  new⭐ 672

This category features evergreen, practical tools for OSINT and security assessments. Their continued popularity is independent of the current AI hype cycle, indicating a stable, ongoing demand for fundamental security utilities.

• soxoj/maigret

  An OSINT tool that collects information on a person by searching their username across thousands of websites. This is a classic, powerful tool for digital investigation that trends periodically due to its broad utility.

  new⭐ 27k
• Z4nzu/hackingtool

  An all-in-one collection of hacking scripts. These 'tool-pack' repositories are common and cater to beginners in cybersecurity, but they typically consist of wrappers around existing, well-known tools rather than novel technology.

  cooling⭐ 73k

A single repository for a financial foundation model highlights the move towards domain-specific AI. Rather than general-purpose models, this indicates a strategic push to train specialized models on proprietary data for high-value industries like finance.

• shiyu-coder/Kronos

  Described as a foundation model for financial markets, this project represents a significant investment in domain-specific AI. Training a large model specifically on financial data aims to capture patterns that general models would miss.

  cooling⭐ 24k