v1.1.8

Added

• Richer generated demo corpus content (theme systems, incidents, actions, risks, and artifacts) so Load demo corpus produces more distinctive, cross-linked documents.
• Batched embedding worker requests that pack many short documents into shared model inference calls instead of one round trip per document.

Changed

• Commit parsed ingest results in fixed-size store/layout batches so large drops update the graph in groups rather than once per file.
• Scale the AI core with corpus size so the center orb stays readable against the expanded force-layout shell.
• Skip IndexedDB original retention for reconstructable generated demo documents, avoiding thousands of redundant writes on demo load.
• Serve and fetch the unversioned demo manifest with no-cache / no-store so desktop upgrades no longer keep a year-old cached corpus definition.