rfauto transforms RF and microwave engineering from manual simulation runs into a fully automated, reproducible pipeline. Integrating 16 simulation engines behind a unified interface—including HFSS, ADS, openEMS, and Elmer—it provides 75 parameterized device templates and over 340 deterministic calculators. The system enforces strict physics-grounded provenance: every dimension, frequency, and metric originates strictly from deterministic kernels or solvers rather than generative hallucinations. With 166 built-in Model Context Protocol (MCP) tools, AI agents can draft recipes, execute parameter sweeps, arbitrate cross-engine results, and inspect quality gates in an automated and safe sandbox.
主要功能
0138 GitHub stars
02Unifies 16 EM, circuit, and multiphysics solvers (such as HFSS, openEMS, and ADS) under a single YAML recipe schema.
03Provides 75 parameterized RF device templates and over 340 deterministic calculators with verifiable formula provenance.
04Exposes 166 MCP tools and resources for AI agent orchestration within a sandboxed validation environment.
05Enforces strict multi-tier quality gates catching mesh artifacts, passivity violations, port semantics, and anchor drift.
06Includes a browser-based local workbench with 24 views for inspecting S-parameters, 3D field visualizations, and optimization runs.
使用案例
01Automating high-frequency passive component synthesis and optimization without manual GUI interactions.
02Enabling AI coding agents to orchestrate complex electromagnetic simulations while ensuring physics results remain deterministic.
03Cross-arbitrating simulation results across commercial and open-source solvers like HFSS and openEMS to validate EM behavior.