BRL-CAD Agentic MCP Interface

BRL-CAD is a battle-tested Constructive Solid Geometry (CSG) engine, but it currently lacks a native, bidirectional interface for headless command execution, creating a massive barrier for modern web integrations and AI agents. This project proposes building BRL-CAD’s first native, extension-agnostic C++ Local IPC (Inter-Process Communication) API. By securely binding a stdio-based pipe listener to the main event loop and leveraging native libbu utilities for memory-safe string parsing, this infrastructure will allow users to safely pipe commands directly into the libged database engine. Designed with a "generic passthrough" architecture, it avoids hardcoding 400+ command schemas and ensures frictionless portability to BRL-CAD’s upcoming MOOSE/Qt interface. Alongside this foundational API, the project delivers a robust Model Context Protocol (MCP) server to establish BRL-CAD's first native Agentic AI integration. Operating externally as decoupled Python-based FastMCP middleware, this daemon bridges the legacy C-engine with modern Large Language Models. Rather than embedding a bulky AI chatbot inside MGED, the user interacts through their preferred external MCP client (like Cursor IDE or Claude Desktop), which routes commands directly into BRL-CAD's new background IPC pipeline. This empowers users to seamlessly query native documentation, automate batch database tasks, and debug spatial relationships using conversational commands, drastically lowering BRL-CAD's onboarding friction.