Context: we use Fellow (meeting notes/transcription) piped into Claude, which creates and updates Asana tasks based on what’s discussed in our meetings (an internal workflow we call FCA — Fellow → Claude → Asana). Part of that workflow includes a pre-flight check where Claude looks for existing tasks that might be relevant to what’s being discussed, even if they weren’t explicitly linked in the meeting notes — currently scoped to a 7-day lookback window.
The problem: that 7-day window is a rough proxy, and it has two failure modes — it can miss genuinely relevant tasks that fall just outside the window, and broader keyword/API-based searches for “similar” tasks can sometimes surface the wrong task entirely, since there’s no guarantee the search matches what was actually reviewed in the meeting.
What would actually solve this: we already maintain curated dynamic saved-search dashboard widgets (completed this week, in progress, new this week, waiting on stakeholder, overdue) that represent exactly the working set our team reviews in each meeting. If Claude (via the Asana MCP integration) could read the current result set of a saved search by its ID — the same way a person viewing that widget would see it — that would give a far more precise, intentional reference set than either the fixed lookback window or a generic search query.
Requested capability: API/MCP-level read access to a saved search’s current results (task list), not just the ability to create/update individual tasks by ID.
This might have been brought up prior to about API access for dynamic widgets with a concern about how the data layer is configured in laymen’s term (from my memory) but wanted to float it up here this is pertinent to anyone else!