STACKBENCH is a multi-agent AI research copilot that evaluates developer frameworks using real GitHub metrics, Wikipedia evidence, and dual-LLM reasoning (Groq + Gemini). It runs agents in parallel, verifies claims, and delivers architect-grade recommendations via Streamlit.

Strategic Analysis: NIST vs. SANS Framework Implementation

框架/代码库主题深度拆解工作流（Claude Code skill）：自顶向下+自底向上双路扫描，关键节点人在回路确认范围，干净上下文子代理交叉验证，每阶段硬退出标准。让你在写技术文章/做分享前，先完整、不漏地理解一个跨模块主题。