nekhss/viyan

Viyan is a project built for an MCP hackathon that models space traffic management for satellites. MCP, or Model Context Protocol, is a standard that lets AI assistants call external tools, and Viyan exposes its satellite safety features as a set of MCP tools so an AI assistant can use them directly. The system tracks active satellites and looks for future close approaches between them, called conjunctions, then assesses how risky each one is. It also uses a machine learning model to predict the probability that a conjunction will actually result in a collision, and it can recommend an automated maneuver a satellite could take to avoid one, described in the project as a negotiation between satellites. The project is split into two main parts. A Python backend, built with the FastAPI framework, handles the underlying simulation, the collision risk calculations, and the machine learning predictions, using libraries such as Skyfield for orbital mechanics, NumPy and Pandas for data handling, and Scikit-learn for the machine learning model. A separate MCP server, written in TypeScript using the NitroStack framework, wraps those backend capabilities as tools, resources, and prompts that an AI assistant can call, including tools to run a simulation, list tracked satellites, check for conjunctions, assess risk, predict collision probability, and request a negotiated avoidance maneuver. To run it, the Python backend is started with a standard FastAPI development server and exposes an interactive API documentation page, while the MCP server is a separate Node.js project installed and run with npm. The README describes this as an early stage hackathon project, and lists planned future work including handling negotiations between more than two satellites at once, streaming live orbital data, real time collision alerts, and an interactive dashboard. The project is released under the MIT license.