mlx-bun library API
    Preparing search index...
    • Parse model key: value lines into a clean info-block field list. Keeps only snake_case keys with a non-empty value (first occurrence wins), drops any model-supplied kind/aliases (we own those) and any physical-spec key that is ungrounded for this entity kind (a world-fact spec key on a domain is a parroted example, never a fact), then prepends type: (if any), the declared kind:, and appends aliases: when we have them. Always returns at least the kind: line, so the article always has a parseable infobox.

      Parameters

      • raw: string
      • kind: "thing" | "person" | "domain" | "project" | "standard"
      • aliases: string[] = []

      Returns string[]