Skip to main content
This page documents the Python MCP surface (mcp.heybtw.com). It reflects mcp-server/tools/cross_event_patterns.py as of this page’s publication — see Schema stability.

Purpose

get_cross_event_patterns groups event attendance by account (domain) and returns each account’s full event history with a computed engagement trajectory — increasing, stable, decreasing, or new. It’s the tool for “who’s compounding engagement across our program,” not just “who came to one event.”

When to use this

  • “Which accounts have attended 3+ events and are still open pipeline?”
  • Identifying accelerating accounts for a targeted follow-up or invite campaign — feed the output into get_account_relationship for the ones worth a deeper look.
  • Feeding get_workspace_overview’s engagement_health block is this tool’s own logic reused at workspace scale — call this tool directly for the account-level detail behind that rollup.

Parameters

Default behavior

Called with no parameters, returns up to 20 accounts with 2 or more distinct events attended, sorted by event count descending then most-recent activity, excluding the workspace’s own domain and any known partner domains. This is the single most consequential default to know: an account with exactly 1 event attended will never appear unless you pass min_events: 1.

Sample request

Sample response

Response field reference

Envelope

data[i]

Errors

Same generic-sanitized-error behavior as every tool on this surface — see get_workspace_overview — Errors. No parameter here can trigger a validation failure; an unmatched domain filter simply returns result_count: 0 with no coverage_note.

Pagination and limits

No cursor-based pagination. limit (default 20, clamped to [1, 500]) caps the number of accounts returned, applied server-side after sorting by event count then recency — there is no way to fetch “the next 20” beyond raising limit. If you need a specific account regardless of rank, use the domain filter or call get_account_relationship directly.