Every couple of weeks, Mark walks through what's happening in the Tana community Slack. Week 10 opens with a roundup of recent releases, then digs into search behavior: applying a search change to every instance, lightweight auto-fill, searching deep under a tag, descriptions in unexpected places, and keeping the current node out of its own results.
Recent releases and community [0:00]
A quick roundup before the questions: the MCP community competition (week three went to a headless-browser setup that drives ChatGPT and captures the results into Tana), a recent Tana Systems Lab on filling out your system with MCP, the new global web clipper and quick capture (shift + control + space), the local API MCP now on by default, and new AI models (GPT 5.4, Opus 4.6, Gemini Pro 3.1).
Push a search node change everywhere [3:10]
Ryan Baum changed a search on one day's page and found it didn't carry to other days. That's intentional: Tana won't overwrite your per-instance view unless you ask it to. Run command + K then update search node on tag and the display, sorting, grouping, and filtering apply to every instance of that tag.
Auto-fill fields on added [4:47]
Marius wanted the mobile voice-capture auto-fill behavior while typing on desktop. Add the auto fill fields command to a tag's on-added automation, and the text you write when you tag a node is used to populate its fields, dates included. Auto-fill is a lightweight starting point; for more accuracy use a run AI fields command instead, or add AI instructions on the individual fields to guide it.
Search descendants under a tag [7:15]
Abby wanted to find text nested any number of levels below a tagged node. A plain string match finds the text everywhere; to scope it, add the path system field with the tag you care about. Path holds every ancestor above a node (the breadcrumb you see above each panel), so the query keeps only matches that sit somewhere under that tag.
Descriptions on tags and fields [9:39]
A hidden-in-plain-sight one: command + K then description adds subtext to a node, but you can also add descriptions to tags and to custom fields (not system ones). The keyboard shortcut is control + I (alt + I on Windows). Handy for documenting what a field is for right where people use it.
Exclude the current node from a search [12:00]
To find related nodes (say, other tasks in the same area) without the current node showing up, use the node name system field with a not clause: not node name parent. That feeds the node you're on into node name and excludes it. The full list of system field names is in the docs, or ask the help chat for them.

