Pinned Loading
-
Visual-Agent-Design
Visual-Agent-Design PublicVisual Agent Design (VAD) — A visual-first framework for AI task diagnosis, routing, Visual Agent Cards, agent execution, validation, and human–AI collaboration.
Python 5
-
VAD-Promptless
VAD-Promptless PublicSkill-first 方法論:Zero Prompting for End Users。把專家方法封裝成可重複、可分享、可版本化的 Skill,並用 VAD 設計 Agent 自主決策。跨模型(ChatGPT/Claude/Gemini CLI)可用。
Python 2
-
ai-to-agent-enterprise-security-os
ai-to-agent-enterprise-security-os Public以 VAD/VAC、分級自主授權與證據閉環,設計由 Agent 承擔日常營運的企業自主資安系統方法論
Python 2
-
-
ai-to-agent-trc-3d
ai-to-agent-trc-3d PublicTRC-3D 人機協作三維決策模型:任務完成、總成本可控,銜接 VAD/VAC(方法論與教學試辦版)
Python 2
-
ai-to-agent-visual-style-agent
ai-to-agent-visual-style-agent PublicVSA視覺設計代理:以VAC、風格規格與驗收流程完成品牌知識圖卡
If the problem persists, check the GitHub status page or contact support.
