1. mvanhorn/last30days-skill Not new today
AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary
First seen: March 24, 2026 | Streak: 3d
Analysis
The `last30days-skill` is an AI agent plugin designed to provide comprehensive, grounded research on any topic by aggregating data from across the web, social media, and prediction markets. By leveraging a multi-signal pipeline, the tool scrapes platforms like Reddit, X, YouTube, Hacker News, and Polymarket, applying a sophisticated scoring system that factors in engagement velocity, source authority, and cross-platform convergence. Technically, the skill utilizes advanced query expansion and deductive reasoning to synthesize findings into a structured narrative, which is then automatically saved as a Markdown file for the user's personal research library.
This tool is highly beneficial for researchers, developers, and trend analysts who need to cut through noise to identify what communities are actually discussing, betting on, or adopting in real-time. It is particularly valuable for those seeking to validate AI prompting techniques, track viral product trends, or gain context on fast-moving cultural and technical shifts. The project is currently gaining traction because it transforms the overwhelming, fragmented nature of internet discourse into actionable, cited insights, effectively acting as an automated intelligence assistant for power users of Claude Code and other CLI-based AI environments.
Teams-first Multi-agent orchestration for Claude Code
First seen: March 26, 2026 | Streak: 1d
Analysis
*Oh-my-claudecode* (OMC) is a sophisticated multi-agent orchestration framework designed to enhance Claude Code by automating complex development workflows through a "team-first" approach. Technically, it functions as a plugin that manages staged pipelines—handling planning, execution, verification, and automated fixing—while leveraging specialized agents and smart model routing to optimize for both performance and token costs. It further extends capabilities by integrating tmux-based CLI workers, allowing users to spawn and orchestrate parallel processes across different models like Codex and Gemini directly within their terminal.
This project is ideal for software engineers and development teams looking to scale their productivity by offloading routine coding, debugging, and architectural tasks to a highly autonomous, multi-model system. It is trending because it drastically lowers the barrier to entry for complex AI-assisted development, offering a "zero-learning-curve" interface that replaces manual prompts with intelligent, persistent, and self-correcting workflows. By turning vague requirements into structured, verifiable code through deep interviewing and automated feedback loops, OMC effectively transforms the command line into a collaborative, multi-agent development studio.
3. virattt/dexter Not new today
An autonomous agent for deep financial research
First seen: February 04, 2026 | Streak: 1d
Analysis
Dexter is an autonomous AI agent designed to perform complex financial research by decomposing abstract queries into structured, actionable research plans. It operates by utilizing real-time market data—such as income statements and cash flow analysis—while employing self-reflection and validation loops to ensure its conclusions are accurate and data-backed. Technically, the system is built on the Bun runtime and integrates with various LLM providers and specialized financial data APIs, logging every reasoning step and tool execution into a local JSONL scratchpad for full transparency and debuggability.
Financial analysts, investors, and researchers would benefit from this project as it automates the tedious, time-consuming process of gathering and synthesizing multi-source financial data. The repository is gaining traction because it provides a modular, "agentic" workflow that significantly reduces the manual effort required for institutional-grade market analysis. By offering features like WhatsApp integration and a robust evaluation suite, Dexter lowers the barrier to entry for users who need reliable, AI-driven insights delivered in a transparent and verifiable manner.
4. ruvnet/RuView Not new today
π RuView: WiFi DensePose turns commodity WiFi signals into real-time human pose estimation, vital sign monitoring, and presence detection — all without a single pixel of video.
First seen: March 03, 2026 | Streak: 2d
Analysis
RuView is an innovative edge AI perception system that enables human pose estimation, vital sign monitoring, and presence detection by analyzing WiFi Channel State Information (CSI) instead of traditional video feeds. By leveraging inexpensive hardware like ESP32-S3 modules, the system processes radio signal disturbances through a high-performance Rust pipeline to reconstruct 3D skeletal movements and biological rhythms in real time. This physics-based approach allows for through-wall sensing and multi-person tracking without the need for cameras, cloud-based data processing, or labeled training sets.
This technology is highly valuable for privacy-conscious organizations, healthcare providers, and disaster response teams who require contactless monitoring in sensitive environments. It is trending because it democratizes sophisticated spatial intelligence, offering a low-cost, secure, and infrastructure-free alternative to traditional surveillance or wearable medical devices. By enabling environments to "see" and interpret human activity through existing radio frequency signatures, RuView represents a significant leap forward for ambient intelligence and accessible, hardware-efficient AI deployment.
5. bytedance/deer-flow Not new today
An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.
First seen: February 26, 2026 | Streak: 4d
Analysis
DeerFlow 2.0 is an open-source, ground-up rewrite of a super-agent harness designed to orchestrate complex, long-running tasks through a combination of sub-agents, long-term memory, and isolated sandboxes. It leverages extensible skills and advanced tools, such as the BytePlus-developed InfoQuest search engine, to research, code, and execute multi-layered workflows. Technically, the framework utilizes a modular architecture that supports various execution environments, including local, Docker, and Kubernetes-based sandboxes, while offering seamless integration with messaging platforms like Slack, Telegram, and Feishu for remote task management.
This project is highly beneficial for developers and AI researchers seeking a robust, scalable infrastructure for building autonomous agents capable of handling intricate, time-consuming operations. Its recent surge in popularity is driven by the 2.0 release's versatility, offering a production-ready solution that bridges the gap between simple chatbots and sophisticated, multi-agent systems. By providing an extensible framework that simplifies complex configurations—such as MCP server support and diverse model compatibility—DeerFlow has established itself as a premier tool for those looking to automate advanced research and development cycles.
No description
First seen: March 26, 2026 | Streak: 1d
Analysis
Insanely Fast Whisper is a highly optimized command-line interface designed to perform rapid audio transcription using OpenAI’s Whisper models on local hardware. By leveraging Hugging Face Transformers, Optimum, and Flash Attention 2, the tool achieves significantly higher throughput than standard implementations through techniques like batch processing, half-precision (fp16) computation, and hardware-specific acceleration. It is specifically built for NVIDIA GPUs and Apple Silicon (via the mps backend), offering advanced features such as speaker diarization and chunked or word-level timestamping.
This project is ideal for developers, researchers, and content creators who need to transcribe large volumes of audio quickly without relying on expensive cloud-based APIs. It is trending because it transforms the state-of-the-art Whisper Large-v3 model into a production-ready, local utility that can process hours of audio in minutes. Its ease of use, combined with transparent benchmarks and community-driven development, makes it a go-to solution for anyone looking to maximize speech-to-text efficiency on their own hardware.
7. agentscope-ai/agentscope Not new today
Build and run agents you can see, understand and trust.
First seen: March 04, 2026 | Streak: 1d
Analysis
AgentScope is a production-ready framework designed for building and orchestrating agentic LLM applications with a focus on reasoning and tool utilization. It offers a comprehensive suite of abstractions—including support for memory, planning, multi-agent message hubs, and Model Context Protocol (MCP) integration—that allow developers to build complex workflows without overly rigid prompt constraints. Technically, the framework is built for extensibility and scale, supporting features like real-time voice interaction, agentic reinforcement learning, and deployment options ranging from local execution to cloud-based Kubernetes clusters.
This project is highly beneficial for AI engineers and developers seeking to build sophisticated, autonomous agent systems that require more than simple prompt chaining. It is currently trending because it addresses the industry's need for a "production-ready" agent stack that combines rapid prototyping with advanced capabilities like human-in-the-loop steering, automated evaluation, and fine-tuning. By providing an open-source, flexible, and well-documented ecosystem, AgentScope empowers users to move beyond experimental chatbots into reliable, multi-agent systems capable of solving complex, real-world tasks.
Building a modern alternative to Salesforce, powered by the community.
First seen: March 26, 2026 | Streak: 1d
Analysis
Twenty is an open-source CRM platform designed as a modern, user-centric alternative to traditional industry leaders like Salesforce. The project offers a highly customizable interface featuring Kanban and table views, granular role-based permissions, and automated workflows, all built on a robust tech stack utilizing TypeScript, NestJS, and PostgreSQL. By prioritizing modularity and extensibility, the platform enables teams to manage customer data, emails, and calendar events within a cohesive, developer-friendly ecosystem.
Organizations looking to escape vendor lock-in and high licensing costs will find Twenty particularly beneficial for its transparent and community-driven development model. The project is trending because it applies contemporary UX design patterns—reminiscent of Notion or Linear—to the often outdated and cumbersome world of enterprise CRM software. Its commitment to open-source principles ensures that companies can self-host their data while contributing to an evolving feature set that addresses the specific needs of modern, agile teams.
OCR model that handles complex tables, forms, handwriting with full layout.
First seen: March 26, 2026 | Streak: 1d
Analysis
Chandra OCR 2 is a state-of-the-art optical character recognition model designed to convert images and PDFs into structured formats like HTML, Markdown, and JSON while maintaining precise layout integrity. The system excels at processing complex elements such as tables, mathematical equations, forms, and handwritten notes across more than 90 languages. Technically, users can deploy the model via a CLI tool using either a lightweight local HuggingFace inference method or a high-performance remote vLLM server for production-scale tasks.
This project is ideal for developers, researchers, and enterprises that require reliable document digitization for data extraction and automation workflows. It is gaining significant traction because it consistently outperforms major industry benchmarks, offering an open-weight alternative that balances high accuracy with flexible deployment options. By providing a specialized solution for complex, multilingual, and semi-structured documents, Chandra 2 effectively addresses common pain points in document intelligence that general-purpose vision models often struggle to resolve.