agent-security-audit

Use when auditing AI agent code for security vulnerabilities — prompt injection detection, confused-deputy attacks, MCP config auditing, tool-chain exfiltration, and OWASP Agentic Top 10 compliance for 刑部 security review. Based on agent-audit (HeadyZhang/agent-audit, 172⭐, 49 rules, 94.6% recall) and prompt-guard (seojoonkim/prompt-guard, 152⭐, 840+ pattern…

Independent GitHub snapshot · Automated classification · Original documentation governs