Hancheng Lou

Hancheng Lou

Interested in code-based cryptography and zero-knowledge proof systems.

zkAgent: Verifiable Agent Execution via One-Shot Complete LLM Inference Proof

Published in Submitted to USENIX Security (under review), 2026

Lizheng Wang, Hancheng Lou, Chongrong Li, Yu Yu, and Yuncong Hu.

This work proves end-to-end LLM agent execution, including complete inference and tool interactions, with one-shot transcript proving for better scalability.