Project config
The metadata that describes the work itself, stored with the project and shared with it.
Project config
What this work is: its medium, language, domain, and the few declared facts the engine should not have to guess.
Most of what the engine knows, it reads from your prose. A handful of things it should never infer, because guessing them wrong quietly distorts everything downstream. Those are declared once, here.
What it holds
| Field | Why it is declared, not inferred |
|---|---|
| Domain | Fiction and non-fiction are read differently. A podcast transcript is not a novel with odd formatting |
| Medium | Screenplay, novel, visual novel, transcript. It changes what the surfaces expect |
| Status | Where the work is in its life, from draft to finished |
| Languages | Which language the prose is in, so entity matching works on the right script |
| Title and logline | The work's own identity, not a filename |
| Genre, themes, keywords | Declared framing, kept separate from anything the engine observed |
Declared values are held apart from observed ones on purpose. When the engine reports that a character drifted from who they were declared to be, that comparison only means something if the two never got merged.
Where it lives
In the project folder, as a plain file beside your prose. It is shared when the folder is shared, and it is readable and editable without the app, like everything else NVS writes.