FluxEDA: A Unified Execution Infrastructure for Stateful Agentic EDA

Zhengrui Chen; Zixuan Song; Yu Li; Qi Sun; Cheng Zhuo

cs.ARMar 26, 2026

FluxEDA: A Unified Execution Infrastructure for Stateful Agentic EDA

Authors: Zhengrui Chen, Zixuan Song, Yu Li, Qi Sun, Cheng Zhuo

Organizations: Zhejiang University, Hangzhou, China

Abstract

Large language models and autonomous agents are increasingly explored for EDA automation, but many existing integrations still rely on script-level or request-level interactions, which makes it difficult to preserve tool state and support iterative optimization in real production-oriented environments. In this work, we present FluxEDA, a unified and stateful infrastructure substrate for agentic EDA. FluxEDA introduces a managed gateway-based execution interface with structured request and response handling. It also maintains persistent backend instances. Together, these features allow upper-layer agents and programmable clients to interact with heterogeneous EDA tools through preserved runtime state, rather than through isolated shell invocations. We evaluate the framework using two representative commercial backend case studies: automated post-route timing ECO and standard-cell sub-library optimization. The results show that FluxEDA can support multi-step analysis and optimization over real tool contexts, including state reuse, rollback, and coordinated iterative execution. These findings suggest that a stateful and governed infrastructure layer is a practical foundation for agent-assisted EDA automation.

Figures & tables

Figure 1. Comparison between traditional EDA workflows and the proposed unified FluxEDA, highlighting the shift from fragmented scripting and tool-specific orchestration to a cohesive, standardized framework for EDA tool integration.

Figure 2. Overall software architecture of FluxEDA. The system organizes heterogeneous EDA environments into a five-layer stack that provides unified access, controlled capability exposure, and persistent tool execution for agentic and programmable workflows.

Figure 3. Execution signal flow in FluxEDA. Agent requests are issued through task logic, forwarded by the RPC client and session manager to the Tcl gateway, resolved through registered api_* methods, and executed on managed EDA runtimes. Capability discovery, method metadata, typed results, artifacts, and execution status traverse the same structured path.

Figure 4. Lifecycle of a managed backend instance in FluxEDA. After a tool process is created and bound to an instance_id , multiple requests can be routed to the same live instance before it is eventually reclaimed by runtime management.

State	Setup WNS (ns)	Setup TNS (ns)	Setup Violating Paths	Hold WHS (ns)	Hold THS (ns)	Hold Violations
Baseline	-0.81	-37.37	140	-0.39	-1.33	4
Iteration 1	-0.76	-34.78	130	-0.38	-1.33	4
Iteration 2	-0.76	-34.84	133	-0.39	-1.33	4
Iteration 3	-0.76	-34.78	130	+0.01	0.00	0

Table 1. Automated post-route timing ECO results. Iteration 2 is rejected because more aggressive setup sizing does not improve WNS and slightly degrades TNS and the number of setup-violating paths. Iteration 3 resumes from the preserved Iteration 1 state and performs targeted hold repair.

State	Area ( $\mu\mathrm{m}^{2}$ )	WNS (ns)	TNS (ns)	Viol. Paths	Used Cell Refs. ( $\downarrow$ vs. Base)	Used Cell Families. ( $\downarrow$ vs. Base)
Baseline	14651.55	-0.246	-20.996	122	149 (0.0%)	51 (0.0%)
A-Run9	12530.89	-1.114	-186.934	237	22 (85.2%)	20 (60.8%)
T-Run04	12748.21	-0.791	-97.780	228	23 (84.6%)	20 (60.8%)
T-Run10	14580.79	-0.385	-41.803	181	30 (79.9%)	20 (60.8%)
T-Run11	13659.95	-0.448	-40.306	157	36 (75.8%)	23 (54.9%)

Table 2. Key milestones in Pareto-driven standard-cell sub-library selection. The last two columns report the number of exact cell references and cell families used in each synthesized implementation.

Figure 5. Search trajectory for Pareto-driven standard-cell sub-library selection. Blue arrows denote the area-driven exploration stage ( A-Runs ) from the minimal library toward low-area operating points. Red arrows denote the subsequent timing-recovery stage ( T-Runs ), where the agent first performs local probing around T-Run04 and then transitions to structure-guided restoration, eventually reaching T-Run11 . Gray dashed arrows indicate alternative local probes.

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

ID	Parent	Change	Decision Intent	Area ( $\mu\mathrm{m}^{2}$ )	WNS (ns)	TNS (ns)	Viol. Paths
Baseline	–	full-library baseline	Reference upper bound with full library support	14651.546	-0.246	-20.996	122
A-Run1	–	minimal textbook X1 seed	Test whether the smallest valid seed library can launch the flow	14064.271	-1.462	-230.246	241
A-Run2	A-Run1	add ND3/NR3 X1	Check whether 3-input logic reduces depth under the minimal seed	13901.532	-1.339	-217.843	241
A-Run3	A-Run2	add AOI12/OAI12 X1	Check whether compound gates further compress critical logic depth	12985.748	-1.301	-208.113	244
A-Run4	A-Run3	add AOI22/OAI22 X1	Complete a basic compound-gate support set and probe the area minimum	12732.037	-1.145	-195.882	244
A-Run5	A-Run4	add OAI12 X2	Test whether endpoint-side complex-gate resizing helps	12763.372	-1.179	-190.739	244

Appendix

Table A1. Complete iteration trajectory for Pareto-driven standard-cell sub-library selection.

Electronic Design Automation

Electronic Design Automation uses algorithms and software to design, verify, optimize, and manufacture electronic circuits and systems.

See all 213 papers

Agentic Electronic Design Automation: A Handoff Perspective
Jun 18, 2026Jiawei Liu, Peiyi Han, Yuntao Lu +3Agent Communication Protocols Electronic Design Automation
The Chinese University of Hong Kong · Primarius Technologies
Can AI Agents Really Complete RTL-to-GDS? Lessons from Benchmarking Tool-Interactive EDA Workflows
Jul 20, 2026Jinyuan Deng, Zhengrui Chen, Xufeng Wei +4Electronic Design Automation LLM Agent Evaluation
1Zhejiang University, China · 2ChipFlux, China
EDATracer: An Agentic Framework for Large-Scale EDA Artifact Analysis
Aug 2, 2026Phat Tieu, Sayanti Jana, Matthew DeLorenzo +5Retrieval-Augmented Generation Electronic Design Automation
Texas A&M University, Department of Electrical and Computer Engineering College Station, Texas 77840, USA

Jun 18, 2026J/K move · Enter open · S save

Agentic Electronic Design Automation: A Handoff Perspective

Jiawei Liu, Peiyi Han, Yuntao Lu, Su Zheng+2

The Chinese University of Hong Kong · Primarius Technologies

Agent Communication Protocols Electronic Design Automation LLM Agent Orchestration

Open paper AskPDF arXiv

Electronic design automation (EDA) is inherently multi-stage and handoff-heavy. Design artifacts, flow scripts, and engineering decisions cross tool, session, and organizational boundaries before final implementation, signoff, or release. Each transfer carries explicit and implicit requirements that may not be fully captured by stage-local checks. LLM-based agents now invoke EDA tools directly, embed retrieved knowledge in executable scripts, and hand off state across sessions and stages. Once their outputs condition downstream engineering decisions, the transferred object must satisfy a handoff contract and meet the assumptions of its next consumer. This survey introduces handoff validity as its organizing principle. A handoff is valid when the transferred object satisfies the consumer's acceptance conditions and carries sufficient context, evidence, and provenance for downstream use. We review 82 systems and classify them into three boundary classes. Stage-Bound systems establish validity within a single EDA stage or bounded verification task. Flow-Bound systems preserve coherent workflow state across tools, invocations, and sessions. Organization-Bound systems maintain source grounding, provenance, scope, and admissibility across knowledge and authority boundaries. For each class, we analyze handoff contracts, handoff objects, coordination mechanisms, and open questions. These analyses motivate a five-layer EDA agent communication protocol (EACP), covering the agent discovery, agent message, tool invocation, workflow orchestration, and security and IP protocols. We aim to provide a common vocabulary and research agenda for trustworthy agentic EDA.

Abstract

Electronic Design Automation

Explore similar work

Agentic Electronic Design Automation: A Handoff Perspective

Can AI Agents Really Complete RTL-to-GDS? Lessons from Benchmarking Tool-Interactive EDA Workflows

EDATracer: An Agentic Framework for Large-Scale EDA Artifact Analysis