Art & Design

Open Agent Harness System for vibe coding, learning, experiment, design and so on.

⚡️ Official docker image for Wallarm Node. API security platform agent.

Reverse-engineer Claude Code sourcemaps to study its architecture, design choices, and reusable AI agent patterns

Explore Claude Code source analysis and AI coding best practices through Harness Engineering, covering architecture, context, permissions, and agent design

Analyze the LangChain DeepAgents source code to master agent harness design, SDK architecture, and production engineering practices through 31 modular chapters.

FCJ Workshop: Build an AI Product Designer Agent on Amazon Bedrock (Claude + Stability Stable Image, AWS CDK, Docker-free)

Higher Order is a plugin package with reusable skills for clearer agent work. It focuses on scoping, success criteria, deterministic writing, project structure, UI design, and explicit option narrowing.

Meta-workspace for Red Hat Cost Management on-prem: git submodules (Koku, koku-ui, cost-onprem-chart, RBAC/Sources UI), a Karpathy-style agent wiki, per-repo constitutions, and a Cursor harness (rules, skills, plans) for Jira-scoped work, OpenShift installs, and UI verification—not a single deployable app.

ICDEV™ v1.2.27 — The System That Builds Systems. AI-powered SDLC with 42-framework compliance automation, 15 AI agents, 11 design canvases, FORGE framework, ANVIL build workflow, and ATO-ready artifact generation.

Official MCP server for bgblur.com — AI background blur & image enhancement tools for Claude and MCP-compatible AI clients.

Find curated frontend agent skills repos with SKILL.md files for easy installation and use across frameworks, UI, testing, and more

Claude Code skill for designing landing pages and portfolios that look human-crafted — custom icons, 2D illustrations, shader effects, no AI slop.

This is a no “vibe coding” zone, my personal AI engineering playground where I learn by building real systems. It focuses on mastering LLMs, RAG, agents, and multi-agent architectures through practical implementation, with an emphasis on clean backend design, modularity, and scalability.

AI-native Mac IDE — Claude agent as first-class Lane, live design surface, deterministic port layout. Private alpha → public OSS.

Computer Use是一个原生 Swift 编写的高性能 macOS 自动化工具集，提供 68+ 条命令 完整覆盖鼠标控制、键盘输入、应用与窗口管理、UI 元素定位、屏幕截图与 OCR、浏览器自动化、文件系统操作、弹窗处理、录制回放、脚本执行、系统通知等全场景计算机控制能力。 工具同时内置 MCP Server 模式，支持AI Agent 直接调用，是构建本地 AI 工作流的理想基础设施。

Agent-first panel backed by a shared data source, with human UI and OpenClaw skill integration.

Turn AI coding agents into research assistants that design, test, and refine experiments with one file

Explore Claude Code design patterns, Agent Runtime, context engineering, and tool systems for building modern AI coding workflows

Self-hosted image bed on Cloudflare Workers + R2 with built-in MCP server for AI agents

Gateway-only Railway image for Hermes Agent with Nous + OpenAI Codex OAuth bootstrap

Community MCP vision bridge for Xiaomi MiMo Vision, enabling image understanding for text-only LLM agents.

Hundreds of agent skills for medical research, including protocol design, data analysis, evidence insights, and academic writing.

Nano Banana Edit — Claude Code skill bridging the Gemini Image API for in-flow image retouching. Shared workspace tool used by /visual-prompt (BIG) and other skills.

Agentic Workflow System — single binary with embedded Studio UI, MCP server, REST API, and React dashboard