工具 API

xmind_get_topic

Get the full rich content of one topic by its id, plus its direct children's ids and titles. Returns the note (plain text in `note`, rich HTML in `noteHtml`), markers, labels, hyperlink, image and task — the values the read markdown only flags the presence of. Task comes back in `task` (null when the topic has none): `status` is the xmind_set_todo checkbox, `progress` / `start` / `due` / `duration` / `priority` the xmind_set_planned_task schedule, in the same units those tools take. When `isRolledUp` is true those values are aggregated from the scheduled descendants rather than set on this topic — report them, but do not write them back or you copy the children's schedule onto the parent. Call it on a child id to walk one level deeper. That is cheaper than xmind_read_mindmap when you want one branch of a large map, but NOT when a node has hundreds of children — check childCount first, and read the whole map instead if this node holds most of it. Start from the rootTopicId that xmind_list_sheets returns. Floating topics come back in `detachedChildren` with their rich content; xmind_read_mindmap names them in its overview too, but only by id and title. Each list is capped at 200: `childCount` / `detachedChildCount` carry the true totals, and `childrenTruncated` / `detachedChildrenTruncated` say which was cut. There is no cursor. A cut `children` list can be read in full with xmind_read_mindmap when this topic is part of the outline; a floating topic's own children never appear there, so keep drilling with this tool.