FEA-MCP Server

The FEA‑MCP Server is a specialized Model Context Protocol (MCP) service that bridges AI assistants with commercial finite‑element analysis (FEA) tools such as ETABS and LUSAS. By exposing a uniform set of MCP endpoints, the server allows an AI to create, query, and manipulate FEA geometry without needing to understand each software’s proprietary API. This abstraction solves the common developer pain point of dealing with multiple, inconsistent interfaces when automating structural design workflows.

At its core, the server implements a small but powerful toolkit of geometry‑creation and retrieval functions. For both ETABS and LUSAS, the AI can batch‑create points, lines, surfaces, and volumes using coordinate data supplied in a single request. It can also fetch the full list of modelled objects, enabling downstream reasoning or visualisation steps. ETABS‑specific tools expose frames and areas, while LUSAS provides lines, surfaces, volumes, sweep operations, and an object‑selection capability. The endpoint gives the model’s measurement system so that AI can adjust inputs accordingly.

Developers benefit from this unified interface in several real‑world scenarios. A design assistant can receive natural‑language prompts like “Create a 10 m span beam between these coordinates” and translate them into the appropriate MCP call, which the server then forwards to the underlying FEA software. When coupled with Claude or other MCP‑enabled assistants, users can iteratively refine a model: the AI suggests geometric modifications, queries existing elements for compliance checks, and even triggers post‑processing steps once the server’s future analysis controls are implemented. The ability to sweep points, lines, and surfaces in LUSAS further expands the creative possibilities for complex 3‑D geometries.

Integration into AI workflows is straightforward: the server runs as a local MCP service, and any MCP‑compatible client can register it via its tool configuration. Because the server encapsulates all COM interactions, developers do not need to manage Windows inter‑process communication or version compatibility themselves. The configuration file lets teams quickly switch between ETABS and LUSAS or upgrade to newer software releases without touching the AI code.

What sets FEA‑MCP apart is its focus on geometry as a first‑class citizen in AI‑driven design. By providing both creation and introspection endpoints, it empowers assistants to not only build models but also reason about them. The planned extensions—material definition, load application, simulation execution, and multi‑coordinate support—will turn the server into a full‑stack FEA automation hub, making it an indispensable tool for engineers who want to embed AI into every stage of the structural design cycle.