What is frankenstein-hospital-system?

Frankenstein is a hackathon project built for the Amrita University MCP Hackathon that lets hospital staff create clinical monitoring rules using plain English instead of writing and reviewing new code. In a real hospital, changing an alarm threshold, such as flagging when a patient's temperature crosses a certain point, normally means a software change and an IT review cycle, which can slow down safety critical decisions. This project turns that process into a natural language request that gets turned into a working monitoring agent almost instantly. Under the hood, the system exposes a set of MCP tools along with simulated FHIR healthcare data, which is a standard format for medical records, covering things like vital signs, prescriptions, and lab results. When someone types a request like an alert rule, it gets parsed and compiled into a schema rule that is saved into a SQLite database as an active running agent. The dashboard includes eighteen interactive widgets built with React and Next.js, showing patient charts, alert boards, and audit logs, and every agent creation or alert action is logged for compliance and accountability tracking. The project is written in TypeScript and built on a framework called NitroStack, and it supports two ways of connecting to it: a standard method used by desktop AI tools like Cursor or Claude Desktop, and a web based method suited for cloud hosting. Setup involves cloning the repository, installing dependencies for both the backend and the widget frontend, then running a development or production build. This project is aimed at developers exploring healthcare focused AI agent tooling, or anyone curious about how natural language could be used to configure clinical monitoring systems. It comes from a hackathon setting, so it should be treated as a demonstration of the idea rather than a system ready for real hospital use. The project is released under the MIT license.