crm_accounts_search
ChatGPTSearch CRM accounts visible to the delegated user and return bounded public business metadata. Returns at most 8 rows per call. The Core handler returns at most eight accounts and does not provide a total. Check partial and completeness in the result before stating totals or calling a list complete.
crm_contacts_search
ChatGPTSearch contacts visible to the current user in the current workspace. Returns at most 8 rows per call. The Core handler caps contacts at eight rows and does not query a total. Check partial and completeness in the result before stating totals or calling a list complete.
crm_deals_history
ChatGPTRead the bounded history projection for one visible deal. Returns at most 10 rows per call. Recent history is intentionally partial and bounded; it is not an audit-log replacement. Check partial and completeness in the result before stating totals or calling a list complete.
crm_deals_search
ChatGPTSearch deals visible to the current user in the current workspace. Returns at most 20 rows per call. The Core handler returns a scoped total and caps rows at twenty; nextCursor is not implemented. Check partial and completeness in the result before stating totals or calling a list complete.
crm_leads_search
ChatGPTSearch leads visible to the delegated user and return only bounded business fields; lead name, email, and phone are excluded. Returns at most 20 rows per call. Scoped total is known; rows are capped at twenty and no cursor is implemented. Check partial and completeness in the result before stating totals or calling a list complete.
crm_pipeline_health
ChatGPTReport visible open-deal pipeline health: counts of stale, stuck and soon-to-close deals with the top 10 of each. Deal values and contact names are excluded; the total value is null when masked. Returns at most 30 rows per call. Summary counts cover every visible open deal; each list is capped at the 10 most urgent entries and partial is true when a list was truncated. Check partial and completeness in the result before stating totals or calling a list complete.
crm_pipeline_summary
ChatGPTReturn the Core-computed pipeline aggregate for the current workspace. The Core handler computes a complete scoped aggregate and excludes Closed Lost. Check partial and completeness in the result before stating totals or calling a list complete.
describe_functions
ChatGPTReturn the full contracts of 1 to 10 visible functions by ID: strict input and output JSON schemas, TypeScript declarations, limits, projected fields and completeness semantics, plus the current catalogVersion. Fails for IDs this connection cannot see or for a catalogVersion that is no longer current.
finance_expenses_search
ChatGPTSearch MinzSpend expense records visible to the verified user. The default mine scope includes self and currently authorized proxy records; all is restricted to users with Spend team-view, control, or admin access. Returns at most 20 rows per call. Count is exact within the authorized scope and filters; rows are bounded to twenty and offset pagination is capped at 10,000. Check partial and completeness in the result before stating totals or calling a list complete.
finance_reports_search
ChatGPTSearch MinzSpend expense settlement and reimbursement reports. These are expense reports, not P&L, general-ledger, or other accounting statements. The default mine scope includes self and currently authorized proxy records; all is restricted to users with Spend team-view, control, or admin access. Returns at most 20 rows per call. Count is exact within the authorized scope and filters; rows are bounded to twenty and offset pagination is capped at 10,000. Reports may contain multiple currencies; monetary totals are omitted. Check partial and completeness in the result before stating totals or calling a list complete.
general_activities_search
ChatGPTSearch bounded activities visible to the delegated user, optionally within a date range and sorted by date. Returns at most 20 rows per call. At most twenty scoped activities are returned; the query matches titles only and activity notes are neither searched nor included. Check partial and completeness in the result before stating totals or calling a list complete.
general_calendar_search
ChatGPTSearch bounded native calendar event rows visible to the delegated user, by explicit from/to or by a range shortcut (today, this_week, next_7_days). Returns at most 50 rows per call. Returns stored CalendarEvent rows capped by limit; recurrence rules are not expanded. A range shortcut is resolved as a half-open window in the given timezone (UTC when omitted) and echoed in window. Check partial and completeness in the result before stating totals or calling a list complete.
general_tasks_search
ChatGPTSearch tasks visible to the delegated user, including personal-task privacy rules, with overdue, deal, contact, due-date range and due-date sort filters. Returns at most 50 rows per call. Follows workspace scope, task sharing and personal-task privacy; at most limit rows are returned. overdue is evaluated against the start of today in the given timezone (UTC when omitted). Check partial and completeness in the result before stating totals or calling a list complete.
search_functions
ChatGPTList the read-only Minz functions this connection may call, filtered by an optional text query over function ID, description and domain (up to 50, default 20). Returns each function's ID, description and contract version plus the current catalogVersion. To read records, call the matching fixed tool: its name is the function ID with dots replaced by underscores.
