Clover: A Neural-Symbolic Agentic Harness with Stochastic Tree-of-Thoughts for Verified RTL Repair

TL;DR

Clover is a neural-symbolic framework that uses a stochastic tree-of-thoughts approach to efficiently repair RTL bugs, outperforming traditional and LLM-based methods in coverage and reliability.

Contribution

It introduces a structured search method with dynamic task dispatching and a stochastic tree-of-thoughts mechanism for improved RTL repair performance.

Findings

Clover fixes 96.8% of bugs within a fixed time limit.

It covers 94% and 63% more bugs than traditional and LLM-based baselines.

Achieves an average pass@1 rate of 87.5%.

Abstract

RTL program repair remains a critical bottleneck in hardware design and verification. Traditional automatic program repair (APR) methods rely on predefined templates and synthesis, limiting their bug coverage. Large language models (LLMs) and coding agents based on them offer flexibility but suffer from randomness and context corruption when handling long RTL code and waveforms. We present Clover, a neural-symbolic agentic harness that orchestrates RTL repair as a structured search over code manipulations to explore a validated solution for the bug. Recognizing that different repair operations favor distinct strategies, Clover dynamically dispatches tasks to specialized LLM agents or symbolic solvers. At its core, Clover introduces stochastic tree-of-thoughts, a test-time scaling mechanism that manages the main agent's context as a search tree, balancing exploration and exploitation for reliable outcomes. An RTL-specific toolbox further empowers agents to interact with the debugging environment. Evaluated on the RTL-repair benchmark, Clover fixes 96.8% of bugs within a fixed time limit, covering 94% and 63% more bugs than both pure traditional and LLM-based baselines, respectively, while achieving an average pass@1 rate of 87.5%, demonstrating high reliability and effectiveness.