AI Infra & Protocols(12)
This category is dominated by the construction of the Claude agent ecosystem. Projects range from full orchestration platforms like ruvnet/ruflo to modular SDKs like browserbase/skills, indicating a shift from experimentation to building production-ready agent infrastructure.
An orchestration platform for Claude, designed for deploying multi-agent systems. Its popularity signals a shift from single-purpose agent scripts to more structured, enterprise-focused agentic workflows, solidifying Claude's position in this emerging market.
A multi-agent framework for financial trading. This project applies the abstract concept of LLM agents to a specific, high-value vertical. It's a practical example of how agent swarms are being explored for complex decision-making tasks.
An established agent project that continues to see interest. Its presence among newer, more specialized tools indicates foundational interest in agent architectures, though it now competes with more focused frameworks like ruvnet/ruflo.
A minimal SDK to add a web browsing tool to Claude agents. This repo is significant because it embodies the 'skills' paradigm: breaking down agent capabilities into small, reusable, and easily integrated components.
An open-source platform for managing coding agents. It aims to integrate agents into developer workflows by providing tools for task assignment and progress tracking, representing an attempt to productize the agent concept for teams.
A tool that allows Claude models to generate n8n automation workflows. This is a very specific integration that showcases how large language models are being used as natural language interfaces for existing low-code platforms.
A middleware interface written in Go to make various web protocols compatible with the DeepSeek API. It's a technical piece of infrastructure indicating efforts to standardize and improve the ecosystem around non-OpenAI/Anthropic models.
A harness for running and evaluating coding agents. Such tools are crucial for benchmarking agent performance and making development more rigorous, moving beyond anecdotal success stories to repeatable, measurable results.
An open-source harness builder for AI coding, focused on making agent behavior deterministic. Its goal of repeatability is a direct response to the often unpredictable nature of LLM outputs, a key challenge for production use.
A collection of pre-defined agents with specific roles, like a 'frontend wizard'. This project represents a higher-level abstraction, packaging agent capabilities into personas rather than raw skills, aiming for easier adoption by non-specialists.
An agent designed to evolve its own skills from a small seed, aiming for full system control with lower token usage. This project explores the frontier of agent self-improvement, a long-term research goal in the field.
An incremental processing engine for long-horizon agents. This addresses a core technical challenge: enabling agents to perform complex, multi-step tasks over long periods without losing context or becoming inefficient.