Connects Claude to Dart AI's task management API with a focus on batch operations that avoid context window bloat. Instead of looping through tasks individually, you use DartQL selectors (SQL-92 WHERE syntax) to update or delete hundreds of tasks in one call. A 50-task update that normally burns 30K tokens takes around 200. Includes dry run modes for batch operations, CSV import with validation, full CRUD for tasks and docs, and search with filters. Useful when you're wrangling large task lists and don't want intermediate JSON responses eating your context. All operations hit production, but deletes are recoverable through Dart's trash system.
claude mcp add --transport stdio standardbeagle-dart-query -- npx -y @standardbeagle/dart-query