precize/Agentic-AI-Top10-Vulnerability

Covers ten distinct threat categories affecting autonomous AI systems—from authorization hijacking and goal manipulation to multi-agent exploitation and supply chain attacks—each with detailed mitigation strategies. Organizes risks around agent-specific attack surfaces (memory manipulation, orchestration loops, temporal attacks) that extend beyond traditional LLM vulnerabilities, mapping them to OWASP frameworks for enterprise security alignment. Developed collaboratively across major cloud and security vendors (Cisco, Palo Alto Networks, Google, Meta) to establish industry-standard baselines for agent deployment security.

175 stars. No commits in the last 6 months.