WorksComputer Science
Run, Read Back, Refine: Observation Fidelity in Execution-Grounded EDA Agents
Abstract
Design engineers lose the most time orchestrating flows and patching the gaps where closed EDA tools refuse to interoperate. We present an execution-grounded LLM agent for PyAether, SKILL, and Tcl that closes a generate-execute-refine loop over unified MCP tools (API retrieval, documentation, sandbox execution) and is already deployed in production EDA flows, with a grounding-gap framework positing that EDA-scripting correctness is bottlenecked by observation fidelity at the L4 (situated) layer, which model scale alone cannot reach. On EDA-Scripting-Bench (449 tasks, three languages, four commercial runtimes), the system reaches 76.5% pooled Pass@1 versus 19.7% (pure LLM): the system-level observation axis spans 56.8 percentage points with the model fixed, exceeding the 23.3-percentage-point backbone range with the harness fixed—two axes answering different questions; observation is an under-invested axis whose returns are measurable and graded.