Archana Kumari

Archana Kumari is an independent security researcher focused on agentic-AI security, authorization integrity, and trust-boundary failures in production AI, workflow, and CI/CD systems. She is a co-discoverer of Approval-Provenance Failure (APF), a resume-time authorization flaw investigated across 43 agentic-AI, durable-workflow, and CI/CD systems, with coordinated disclosures involving Microsoft, OpenAI, AWS, and other vendors. Her research examines how persisted workflow state can be mistaken for legitimate authorization and how provenance-bound enforcement can restore trustworthy oversight. She has spoken at Linux Foundation MCP Dev Summit events in Bengaluru and Mumbai and works on practical security architectures for autonomous and agentic systems.