How Indexing Works

Skald maintains the knowledge graph by extracting relationships from your project content and storing them as indexed edges.

What gets indexed

Skald scans:

Conversation descriptions — @-mentions in the conversation summary
Dialogue nodes — @-mentions in dialogue text and assigned characters (conversation participant edges)
Player choice nodes — @-mentions in choice text
Character descriptions — @-mentions referencing other entities
Lore descriptions — @-mentions referencing other entities

Mention extraction

Mentions are parsed from rich text using Skald's language. Only properly typed @-mentions resolve to characters, conversations, or lore items. See Mentions in the Skald language documentation.

In addition to @-mentions, Skald creates Conversation Participant edges when a dialogue node has a character assigned in the inspector. This links the character node to the conversation node even when the character is not @-mentioned in the dialogue text.

Keep mentions up to date

If you rename or delete a character or lore item, @-mentions in rich text fields update automatically. Plain-text references elsewhere in your project are not reindexed — prefer @-mentions for anything you want reflected in the knowledge graph.

How Indexing Works