Hackers Target LiteLLM, RAGFlow and Kestra AI Infrastructure to Steal API Keys and Mine Crypto

Microsoft observed coordinated attacks against AI infrastructure components—LiteLLM, RAGFlow, and Kestra—where attackers exploited CVE-class flaws and misconfigurations to harvest environment variables and API keys, exfiltrate LLM provider credentials via hidden hooks, access databases, deploy XMRig cryptominers, and achieve persistent access; the report includes CVE references and IoCs (IP addresses and ports) to aid detection and response.