Project data and versioning

When you run medatarun in a directory, that directory becomes the project directory. Medatarun creates a subdirectory named .medatarun. It contains the data used by Medatarun and its extensions. Each extension can have its own subdirectory inside it. This is the Medatarun directory.

.medatarun stores your models and any data managed by the extensions. If the project directory is under version control (for example with Git), you should commit .medatarun.

Medatarun is designed to let AI agents manage your models. Committing .medatarun lets you see what these agents change over time and inspect it with standard version-control tools.