Electronic design automation (EDA) is inherently multi-stage and handoff-heavy, with design artifacts, flow scripts, and engineering decisions crossing tool, session, and organizational boundaries. Each transfer carries explicit and implicit requirements that stage-local checks may not fully capture, creating risk when LLM-based agents invoke EDA tools directly, embed retrieved knowledge in executable scripts, and hand off state across sessions and stages. A new survey from researchers Liu, Jiawei, Han, Peiyi, Lu, Yuntao, Zheng, Su, Yan, Fengyu, and Bei introduces handoff validity as the 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.
Three Boundary Classes of Agentic EDA Systems
The survey reviews 82 systems and classifies them into three boundary classes based on where handoffs occur:
| Boundary Class | Scope | Key Concern |
|---|---|---|
| Stage-Bound | Within a single EDA stage or bounded verification task | Validity within a stage |
| Flow-Bound | Across tools, invocations, and sessions | Coherent workflow state preservation |
| Organization-Bound | Across knowledge and authority boundaries | Source grounding, provenance, scope, admissibility |
- 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, the authors analyze handoff contracts, handoff objects, coordination mechanisms, and open questions.
Five-Layer EDA Agent Communication Protocol (EACP)
To address the challenges identified across the three boundary classes, the paper motivates a five-layer EDA agent communication protocol (EACP). The protocol layers are:
- Agent discovery – enabling agents to find each other.
- Agent message – standardizing communication between agents.
- Tool invocation – defining how agents invoke EDA tools.
- Workflow orchestration – coordinating multi-agent workflows.
- Security and IP protocols – protecting intellectual property and ensuring secure handoffs.
The EACP aims to provide a common vocabulary and research agenda for trustworthy agentic EDA, ensuring that handoffs between agents maintain validity across increasingly complex boundaries.
Implications for Enterprise Decision-Makers
For technology leaders evaluating agentic AI in hardware design, the handoff validity framework offers a structured way to assess where failures occur and what contractual obligations transferred artifacts must satisfy. The three-class taxonomy helps identify whether a given agentic EDA solution is designed for stage-internal tasks, cross-tool workflows, or organization-spanning collaboration. The proposed EACP, while still a research agenda, points toward the need for standardized protocols to ensure trust and reliability in multi-agent EDA systems.
