...An AI coding agent or an LLM server writes them out as ready-to-import VTT documents
5. Acceptance test on a staging VTT
6. Import into production and play
Packages move through status folders in the content repo, tracked by Subversion, Git or none. Configuration lives there too, so everyone follows the same workflow.
Supported: Foundry VTT on Linux; other platforms best effort. Requires Python, the Foundry CLI and an OpenAI-ready agent or a llama.cpp server.
See Wiki (https://sourceforge.net/p/vttloremill/wiki/) for more.