Skip to main content
Read-only.
The marketer’s question is “what is left to do for this event?” Reach for this tool whenever the user asks what their follow-ups are, what is overdue, or what a partner still owes. Never rebuild the queue from chat history or from rosters. Follows the heyBTW MCP rules.

When to use it

  • “What’s still open from the Austin dinner?”
  • “What’s overdue for me?”
  • “What does Brightpath still owe us for the open house?”
  • “Show me everything Tony finished last month.”

Parameters

All optional. No limit and no cursor: narrow with the filters instead. At most 500 tasks are read per call; a longer queue says so in a gap.

Response

data is one row per task, ordered by due_date (undated last), then creation time, then task_id, as metadata.ordering states. A host sees every partner assigned; a partner sees only its own assignment, never the host’s own tasks, owner emails or who accepted. A real response from the heyBTW workspace’s ScaleUp.ai fixtures, trimmed to one of four rows and to the coverage notes that matter:

Rules the agent must follow

  • heyBTW decides overdue, not you. A task is overdue when its due_date is before the collab’s local date at metadata.scope.as_of. Never recompute it.
  • Never call a task with a due_gap on time or overdue: it has no due date.
  • A task is a commitment, not evidence of contact. For contact use get_attendee_touch_status, and report the two as separate facts.
  • Tasks on an archived collab leave the open and overdue queue until it is restored; relay the coverage note that counts them.
  • Relay coverage_notes as written.

Errors and gaps

A failed read is always an error, never an empty list.