You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
tana_semantic_search crashed on nodes with a null name (null is not an object (evaluating 'name.includes')). enrichSearchResults copied the nullable nodes.name column straight into the result, then isReferenceSyntax() called .includes() on it — so any result set containing an unnamed source/container node failed the entire search (retrying with a different query sometimes dodged the bad node). Fixed at the source (coalesce null name → "" during enrichment) and hardened isReferenceSyntax() to treat null/undefined/empty as non-reference. The shared enrichment/filter (src/embeddings/search-filter.ts) is used by the MCP tana_semantic_search tool and the CLI (search, embed), so both are covered. Also guarded the webhook server's semantic-result renderer (convertSemanticResultsToTana), which had the same unguarded .includes on a nullable node/ancestor name. (Reported via Claude Code.)