Trace2Skill: Verifier-Guided Skill Evolution for Long-Context EDA Agents

Zijian Du, Nathaniel Pinckney

arXiv:2605.21810 · 2026-05-23 공개 · arXiv · PDF

long-context test-time-scaling skill-evolution verifier-guided verilog-design dense-verifier-feedback hardware-llm trace2skill

Abstract

Complex Verilog Design Problems (CVDP) challenge hardware LLM agents because solving them requires localizing verifier-relevant RTL, testbenches, include paths, and build dependencies inside large repository snapshots, making precise edits, and recovering from sparse hidden-verifier failures. We present Trace2Skill, a test-time scaling framework that improves a hardware agent without RTL-specialized model fine-tuning. Rather than training a new model or only sampling more candidate solutions, Trace2Skill treats the agent's natural-language skill as an evolvable policy. It mines repeated rollout traces for success and failure modes, converts them into dense diagnostics and oracle lessons, and uses an oracle, mutator, and selector loop to produce task-specific skills that guide later search, editing, validation, and recovery. Because final pass/fail labels are often too coarse for hard failures, Trace2Skill also supports bounded runtime dense verifier feedback that returns sanitized functional observations while keeping hidden harnesses and reference solutions inaccessible to the agent. This feedback helps guide skill evolution and agent execution by connecting skill text, verifier evidence, and downstream behavior. Across hard CVDP tasks that defeat the seed CVDP agent, including tasks that also defeat frontier coding agents, Trace2Skill with dense verifier feedback substantially improves task pass rates and produces breakthrough passes on previously unsolved tasks, without requiring high-quality fine-tuning data, specialized RTL model training, or model weight updates. The same framework provides a general test-time scaling strategy that can extend beyond digital design to other verifiable EDA tasks.

한국어 요약

한 줄 요약

**[EDA 에이전트 / Skill Evolution]** Trace2Skill이 복잡 Verilog 설계 문제(CVDP)에서 RTL 특화 fine-tuning 없이 rollout trace 채굴·oracle/mutator/selector 루프로 skill을 진화시키고, dense verifier feedback과 결합해 frontier 코딩 에이전트도 못 푼 task에 breakthrough pass 달성.

핵심 기여도

핵심 아이디어

하드웨어 에이전트의 성능 향상은 모델 가중치 갱신 대신 자연어 skill을 evolvable policy로 진화시키는 것으로 달성 가능하며, rollout trace의 성공/실패 패턴 채굴과 dense verifier feedback의 결합이 sparse signal 문제를 우회한다.

기술적 접근법

주요 결과

의의 및 한계

**의의**: 하드웨어 에이전트 개선을 모델 학습 부담 없이 가능하게 함, skill을 명시적 evolvable artifact로 다루는 패러다임, dense verifier feedback의 보안·실용 균형. **한계**: CVDP·verifiable EDA에 한정, runtime verifier feedback 인프라 비용, skill 진화 루프의 수렴 보장 부재.

실용적 활용