Carsten Rösnick-NeugebauerIf you’re building an #MCPServer, what are the good design practices for tools that might return large datasets? Pagination? Specialized tools with presets like `list_todays_tasks`?
Think I got it: resource/list can be paginated. So provide tasks as resources with `resources/read` on `todoist://tasks` returning all paginated tasks. In addition, provide a `list_tasks(filter)` tool to filter tasks with a limit on the returned dataset size.