VeriTrace: Human-Like Temporal Exploration Completes Agentic Action Space

2026-08-03

AI Summary

Existing approaches to automatic Verilog RTL generation are constrained by a narrow debugging action space, limiting their ability to perform hypothesis-driven root cause analysis and capping accuracy at approximately 95%. This work proposes a novel multi-agent system in which an Inspector agent jointly controls signal selection, temporal windows, and iterative depth, thereby enabling unrestricted exploration across signal, time, and iteration dimensions for the first time. By elevating the debugging paradigm from pattern matching to hypothesis-verification-based root cause analysis, the system—built upon Claude Sonnet 4.0 and integrating a programmable waveform query interface with a dynamic hypothesis generation–validation loop—achieves 100% Pass@1 on the VerilogEval-V2 benchmark. This marks the first method to attain perfect functional correctness, surpassing the strongest baseline by 5.1%.

Abstract

Large language models have shown promise for automated Verilog RTL generation, yet state-of-the-art multi-agent systems plateau at ~95% accuracy on standard benchmarks. We trace this ceiling to an incomplete debugging action space: existing systems restrict which signals the agent can inspect, which time windows it can query, or both, reducing debugging to pattern matching on a narrow, predetermined view of circuit behavior rather than hypothesis-driven root-cause analysis. We present VeriTrace, a multi-agent system whose Inspector agent operates over a complete debugging action space, with independent control over signal selection, time-window bounds, and iteration depth. This capability, which we term Agentic Temporal Exploration, enables the agent to form hypotheses about failure causes, query the waveform for evidence, and refine its understanding iteratively, mirroring the exploratory process of human verification engineers. VeriTrace achieves 100\% Pass@1 on VerilogEval-V2, the first system to attain perfect functional correctness on this benchmark. On a shared Claude Sonnet 4.0 backbone, VeriTrace outperforms the strongest reproduced baseline by +5.1%, demonstrating that debugging agency closes the final accuracy gap.

Problem

• debugging action space
• temporal exploration
• Verilog RTL generation
• multi-agent systems
• root-cause analysis

Innovation

• Agentic Temporal Exploration
• complete debugging action space
• multi-agent verification
• hypothesis-driven debugging
• Verilog RTL generation