AI application compliance gateway · Check an AI project with one command for data export, hardcoded secrets, and personal information exposure (Cybersecurity Law, PIPL, Classified Protection 2.0, data export, AI labeling), and provide suggestions for domestic model alternatives; can also provide runtime protection against prompt injection and data exfiltration · Chinese-first · zero dependencies · open source
MCP server Claude CodeCodexCursor +2
AI agent security: 7 MCP tools for injection detection, PII scanning, command safety, DLP. Runs locally from the shellward npm package. Needs 3 environment variables to run.