GitHub 2026-03-23 Weekly Trending Projects

Native HUD for Claude Code that visualizes context, tools and agents for terminal debugging.

MiroFish is a multi-agent, long-memory simulation prediction platform that maps real-world seed data into high-fidelity digital worlds, enabling natural-language-driven scenario simulations and automated report generation for public opinion, policy and creative sandboxing, while requiring attention to licensing and operational risks.

Superpowers provides a composable skills library and TDD-enforced agent workflow for Claude Code, enabling end-to-end automation from design and planning to subagent execution, with plugin integration and extensibility.

DeepAgents is a Python toolkit for building layered, extensible deep agents that combine planning, subagents, filesystem access and persistent memory to handle long, complex tasks and research workflows; however, repository metadata (license, contributors, releases) is incomplete, so evaluate compliance and stability before production use.

Everything Claude Code delivers a production-grade toolchain for agent harness performance and governance—covering skills, hooks, memory persistence, and security scanning—suited for teams requiring cross-platform parity and measurable quality gates.

MoneyPrinterV2 is an automation toolkit combining social bots, short-video publishing and affiliate promotion, intended for users with Python skills to deploy and customize.

Lightpanda is an ultra-light headless browser (CDP-compatible) focused on low memory and fast execution for large-scale scraping, LLM training and automation.

A progressive, hands-on tutorial repository for developers and researchers that demonstrates how to build Claude-like AI coding agents across five increasing-complexity examples, emphasizing clear concepts and runnable samples.

Vite+ is an integrated CLI toolchain for front-end development that consolidates Vite, testing, formatting, linting and monorepo task management under a single config—suited for teams seeking unified workflows and faster local/CI pipelines.

Unsloth delivers a memory-friendly fine-tuning and RL toolchain for research and engineering, supporting multiple models and free notebooks—great for low-cost prototyping and experiments, but production readiness and performance reproduction require careful evaluation.

OpenViking abstracts agent context as a manageable 'filesystem', offering tiered loading, directory-recursive retrieval and visualization to reduce retrieval cost and improve observability—suitable for AI agent systems requiring long-term memory and complex context management.

TradingAgents constructs a modular, configurable research pipeline of role-based LLM agents, supporting multi-provider models and simulated execution—suited for academic and R&D strategy validation but not investment advice.

shadPS4 is a C++ cross-platform PS4 emulator core with a custom shader compiler and AMD GPU emulation; it suits developers and researchers for compatibility testing and graphics research but remains early-stage, requires user-provided firmware, and has a high usage barrier.

Context Hub: Provides Markdown‑maintained, versioned API docs and local annotation/feedback loops for coding agents, improving code-generation accuracy and continual learning.