* Pure math-based chunking: Split entire content into equal-sized chunks with overlap * Ignores section headers, just splits by size

* Simple chunking: Split by ## headers only

* Close the chunking modal

* Convert keywords to regex patterns ONLY when we find actual related words * Strategy: * 1. Find words in the list that share a common prefix (fu

* Create the modal HTML if it doesn't exist

* Adds keyword-based fallback chunks when semantic search misses. * @param {string[]} queryKeywords Keywords extracted from the query text * @para

* Find a template that matches a lorebook entry * Checks template name, triggers, and categories

* Force scan lorebooks

* Get ALL contextual libraries relevant to the current chat context * Returns: { global: {...}, character: {...}, chat: {...} } with actual library

* Handle RAG button click - vectorize the fullsheet * * @param {number} messageId - Message ID * @param {Object} fullsheetInfo - Fullsheet infor

* Handle weight badge click - shared function for all weight badge clicks

* Hook into ST's message processing to filter BunnyMoTags from AI context * This needs to be called before messages are sent to the AI

* Inject RAG results into AI context * * @param {string} characterName - Character name * @param {Array} results - RAG query results

* Pretty print object data