blackzhanzhan/Cyber-Ming-Protocol

Implements a human-centered governance protocol that routes AI code execution and auditing through separate, isolated channels—an IDE executor submits atomic execution contracts first, a Web auditor independently reviews plans without seeing implementation details, and a human decision-maker approves or rejects each phase. Built for detecting "pseudo-completion" (code that appears done but lacks physical evidence) and cognitive debt accumulation, it uses "seven-star lamp resurrection" techniques for context window recovery and pulse-based workload distribution to maintain high governance without sacrificing throughput.