# `MnemosyneZvex.Schema`

Builds the per-repo `Zvex.Collection.Schema` used by `MnemosyneZvex.Backend`
and centralises atom->string conversion for the `node_type` column.

# `build`

```elixir
@spec build(keyword()) :: Zvex.Collection.Schema.t()
```

Builds a Zvex collection schema from backend opts.

Required: `:dimension`. Optional: `:index` (default `:hnsw`), `:metric`
(default `:cosine`), `:index_opts` (default `[m: 16, ef_construction: 200]`).

# `node_type_string`

```elixir
@spec node_type_string(atom()) :: String.t()
```

Stringifies a node-type atom for the `node_type` zvex column. Used by both
the encoder (writes) and the filter builder (reads) so the two sites cannot
diverge.

# `node_types`

```elixir
@spec node_types() :: [atom()]
```

Returns the canonical list of node-type atoms supported by the backend.

# `version`

```elixir
@spec version() :: pos_integer()
```

Schema layout version. Bumped if breaking column changes are introduced.

---

*Consult [api-reference.md](api-reference.md) for complete listing*