Tana Live Search Deep-Dive: PARENTS and DESCENDANTS

Tana’s live search becomes far more powerful when it’s paired with the parent and descendants operators, letting users pull content either “across the whole workspace” (pull search) or “up from a specific branch” (bubble search). In practice, this means meeting templates can automatically compile agenda items and meeting summaries without manually linking every note—while still respecting where items live in the workspace hierarchy.

The walkthrough builds a meeting-focused super tag system: a meeting tag for the meeting node itself, an agenda item tag extended into agenda items, plus tags for highlights, follow-ups, and resources, and a members tag for collaborative teams. A demo meeting (“design review of cortexfutura.com”) contains search nodes for an agenda and a summary, along with notes created during the meeting. The agenda search is treated as a pull search: it gathers agenda items from around the graph by filtering for items tagged as agenda items that are not done, then using the parent operator to ensure those items are tied back to the meeting node they’re nested under. The key detail is that “parent” lets the search reference the node that the search is indented beneath—so agenda items added on different days and in different parts of the graph still appear in the meeting’s agenda as long as their meeting field points to that parent meeting.

That same setup supports a workflow shortcut: once agenda items are pulled in, users can quickly edit or add items directly from the search results (via the UI affordance shown as an edit action), and the meeting agenda can include candidates that aren’t explicitly connected yet—by unreferencing items and selectively collecting what should be discussed.

For the summary, the video shifts to bubble search. Instead of pulling agenda items from everywhere, the summary search bubbles up only the highlights, follow-ups, and resources that exist under the meeting node’s subtree. This is done with parent and descendants logic: the search walks the descendants of the meeting node and collects matching tags wherever they’re indented within that branch. An important nuance appears when items are added outside the expected indentation: even if a highlight is added at a deeper or different level, it still shows up in the summary as long as it’s a descendant of the meeting node.

The transcript also flags a limitation: parent/descendants searches do not include referenced content (like items linked via references/Wiki-style links) by default. To include those referenced items, the search must append a refs modifier (described as adding “refs” at the end of the search).

Finally, the hierarchy matters when search nodes sit inside fields. When agenda and summary searches are moved into fields that are themselves nested under the meeting node, parent/descendants may stop returning results because the search is now indented two levels deeper. Switching from parent/descendants to grandparent/descendants restores the expected behavior. Adding refs again brings referenced Wiki items back into the results. The net effect: meeting pages can stay clean and automated while still controlling exactly what content is included, where it comes from, and whether references are respected.