TraceCoder: A Trace-Driven Multi-Agent Framework for Automated Debugging of LLM-Generated Code
Published in ICSE 2026, 2026
Recommended citation: Jiangping Huang, Wenguang Ye, Weisong Sun, Jian Zhang, Mingyue Zhang, and Yang Liu. "TraceCoder: A Trace-Driven Multi-Agent Framework for Automated Debugging of LLM-Generated Code." ICSE 2026.
TraceCoder studies automated debugging and repair for LLM-generated code through runtime traces, multi-agent collaboration, causal analysis, historical lesson learning, and rollback-based iterative repair.
Tags: CCF A, Top-tier SE Conference, LLM-generated code debugging, trace-driven multi-agent repair.