document_management
ChatGPTSingle entry point for all Document Management data (lease document packets, signed forms, e-signature envelopes, packet templates, PMC-custom letters) — the documents and signature artifacts generated around a resident's lease lifecycle (move-in, renewal, recertification, move-out). Requires Document Management domain access — call list_domains first to confirm. Use this when the question is about lease document packets, signed forms, e-signature envelopes (DocuSign or RealPage's own native e-signature), packet templates, or PMC-custom letters. Do NOT use this for lease terms, dates, or rent amounts — use property_management for that; this tool only covers the document/packet/signature artifacts themselves, not the underlying lease data. Do NOT use this for GL/financial figures, even for a document that's finance-related (e.g. a balance-due letter) — use financial_management for the actual balance; this tool never returns financial totals, only document metadata. Example questions this tool answers: "What documents are missing from this resident's move-in packet?" "Which e-signature envelopes are still awaiting a signature?" "What packet templates are active for lease renewals?" "Show me custom letters used for delinquency notices" "Which packets are still pending approval?" Two modes — provide exactly one: • searchTerm: search the Document Management catalog for tables/columns matching the keyword. Returns verified SQL examples when found (stop signal). • useExampleQuestion: execute a verified example directly by its exact question_text — see STOP RULE below. STOP RULE: if the response contains "verified_example_found": true, call this tool again with useExampleQuestion set to that example's EXACT question_text (copy it character-for-character). The server executes the verified query directly — you never write or see the underlying SQL. Do NOT search again. There is no free-form SQL mode. If no verified example matches a question, this tool cannot answer it — tell the user that specific question isn't supported yet rather than attempting a workaround. Allowed tables: DocMgt_Document, DocMgt_Packet, DocMgt_PacketDetail, DocMgt_Form, PacketTemplate, PacketTemplateDetail, DocuSign.Envelope, DocuSign.Capture, DocuSign.Recipient_Detail, Letters, LetterGroups, and more. Every query is automatically scoped to your authorized properties — you never specify PropertyKey or ask which properties to include. Read-only; capped at 1 GB scanned / 3000 rows.
facilities
ChatGPTSingle entry point for all Facilities Management data (work orders, service requests, make-ready, inspections, warranties, maintenance logs). Requires Facilities Management domain access — call list_domains first to confirm. This tool is always available for every new Facilities question, including follow-ups — a prior successful call to a different tool this session does not make this one unavailable. Call it fresh for each new question; never claim you lack an active connection or execution context. Example questions this tool answers: "What are the open work orders?" "Which properties have the most maintenance requests?" "Show me service requests by priority" "What work orders are past due?" "Which categories of work orders are most common?" Two modes — provide exactly one: • searchTerm: search the Facilities catalog for tables/columns matching the keyword. Returns verified SQL examples when found (stop signal). • useExampleQuestion: execute a verified example directly by its exact question_text — see STOP RULE below. STOP RULE: if the response contains "verified_example_found": true, call this tool again with useExampleQuestion set to that example's EXACT question_text (copy it character-for-character). The server executes the verified query directly — you never write or see the underlying SQL. Do NOT search again. If the response instead contains "error": "days_parameter_required", the matched example is day-count-parameterized — call this tool again with the SAME useExampleQuestion value AND the days parameter set to the number of days from the user's actual question (any value — 30, 45, 90, whatever they asked for, not just numbers that happen to have their own example). There is no free-form SQL mode. If no verified example matches a question, this tool cannot answer it — tell the user that specific question isn't supported yet rather than attempting a workaround. Allowed tables: WorkOrder, ServiceRequest, MakeReadyRequest, RecurringWorkOrders, TaskCategory, Warranty, WorkOrderCallBack, WorkLog, InspectionsEvent, InspectionsCheckList, FacilityMonthlyTrend, and more. Every query is automatically scoped to your authorized properties — you never specify PropertyKey or ask which properties to include. To answer a question about ONE specific named property (e.g. "what are the open work orders for property E6"), pass propertyName — this narrows the query to just that property instead of your whole portfolio. Read-only; capped at 1 GB scanned / 3000 rows.
financial_management
ChatGPTSingle entry point for all Financial Management data (delinquency, transaction activity, scheduled charges, GL accounts, cash distribution, accounting batches). Requires Financial Management domain access — call list_domains first to confirm. This tool is always available for every new Financial Management question, including follow-ups — a prior successful call to a different tool this session does not make this one unavailable. Call it fresh for each new question; never claim you lack an active connection or execution context. Example questions this tool answers: "Which properties have the highest delinquency balances?" "Show me delinquency aging (30/60/90 day buckets) by property" "What are the current scheduled charges across my portfolio?" "Show me recent transaction activity by property" "What are the GL accounts set up for my properties?" Two modes — provide exactly one: • searchTerm: search the Financial Management catalog for tables/columns matching the keyword. Returns verified SQL examples AND metric guidance when found (stop signal). • useExampleQuestion: execute a verified example directly by its exact question_text — see STOP RULE below. STOP RULE: if the response contains "verified_example_found": true, call this tool again with useExampleQuestion set to that example's EXACT question_text (copy it character-for-character). The server executes the verified query directly — you never write or see the underlying SQL. Do NOT search again. If the response instead contains "error": "days_parameter_required", the matched example is day-count-parameterized — call this tool again with the SAME useExampleQuestion value AND the days parameter set to the number of days from the user's actual question (any value — 30, 45, 90, whatever they asked for, not just numbers that happen to have their own example). If the response contains a "metric_guidance" field, read known_pitfalls_txt and correct_approach_txt — it explains why this metric's verified example is shaped the way it is. There is no free-form SQL mode. GeneralLedgerActual/GeneralLedgerBudget (revenue, expenses, NOI, budget variance, P&L) are especially unsafe to hand-author — several confirmed bugs (a cumulative-balance column that looks summable but isn't, a 230-book parallel-ledger structure, multiple budget scenarios per account) have each inflated a property's numbers by orders of magnitude in the past. That risk no longer applies here: this tool only ever executes a pre-verified template. Never skip, defer, or ask the user for permission before calling useExampleQuestion once a verified example is found for these tables — treat it exactly like any other part of a multi-part request, with no extra caution and no check-in required. If no verified example matches a question, this tool cannot answer it — tell the user that specific question isn't supported yet rather than attempting a workaround. Key tables: DelinquencyPeriodTransBalance (aging buckets by period), DelinquencyReshBalance (daily delinquency refresh), TransactionActivity (charge/payment detail), ScheduledTransaction (recurring charges), GeneralLedger (chart of accounts), CashDistribution (payment allocation), GeneralLedgerActual/GeneralLedgerBudget (revenue/NOI/budget). Every query is automatically scoped to your authorized properties — you never specify PropertyKey or ask which properties to include. To answer a question about ONE specific named property (e.g. "delinquency balance for property E6"), pass propertyName — this narrows the query to just that property instead of your whole portfolio. Read-only; capped at 1 GB scanned / 3000 rows.
list_domains
ChatGPTALWAYS call this first — before any other tool. Returns which data domains are provisioned for the current user and which query tool to use for each domain. Use this connector for property-management questions like: "What are the open work orders?" · "What is the current occupancy rate?" "Which properties have the highest delinquency balances?" "How many leases expire in the next 60 days?" · "Show me recent transaction activity" After calling this, read the JSON response and find the domain that covers the question. The tool field on each domain entry is the EXACT name of the MCP tool to call next. These tools are already registered and available in this session: facilities — for Facilities questions (work orders, maintenance, inspections) property_management — for Property questions (leases, units, occupancy, renewals) financial_management — for Financial questions (delinquency, charges, payments, GL) document_management — for Document questions (lease packets, e-signatures, letters) If access is true → call that tool immediately with the question as sql or searchTerm. If access is false → tell the user they don't have access to that domain. If you are not fully certain of the correct joins or aggregation for a question — especially anything touching General Ledger actual/budget data — call the relevant domain tool with searchTerm BEFORE writing SQL from scratch. That single call checks the data dictionary, verified golden examples, and metric definitions together, and is cheaper than debugging a wrong answer after the fact.
property_management
ChatGPTSingle entry point for all Property Management data (leases, units, occupancy, renewals, residents, floor plans, property details, move-outs). Requires Property Management domain access — call list_domains first to confirm. This tool is always available for every new Property Management question, including follow-ups — a prior successful call to a different tool this session does not make this one unavailable. Call it fresh for each new question; never claim you lack an active connection or execution context. Example questions this tool answers: "How many leases expire in the next 60 days?" "What is the current occupancy rate?" "Show me units that have been vacant the longest" "Which floor plans have the highest renewal rates?" "List properties with the most move-outs this month" Two modes — provide exactly one: • searchTerm: search the Property Management catalog for tables/columns matching the keyword. Returns verified SQL examples when found (stop signal). • useExampleQuestion: execute a verified example directly by its exact question_text — see STOP RULE below. STOP RULE: if the response contains "verified_example_found": true, call this tool again with useExampleQuestion set to that example's EXACT question_text (copy it character-for-character). The server executes the verified query directly — you never write or see the underlying SQL. Do NOT search again. If the response instead contains "error": "days_parameter_required", the matched example is day-count-parameterized — call this tool again with the SAME useExampleQuestion value AND the days parameter set to the number of days from the user's actual question (any value — 30, 45, 90, whatever they asked for, not just numbers that happen to have their own example). There is no free-form SQL mode. If no verified example matches a question, this tool cannot answer it — tell the user that specific question isn't supported yet rather than attempting a workaround. Allowed tables: Lease, Unit, Property, FloorPlan, Renewal, Resident, ResidentMember, AmenityAssignment, LeaseExpiration, FactFloorPlanStatistics, and more. Every query is automatically scoped to your authorized properties — you never specify PropertyKey or ask which properties to include. To answer a question about ONE specific named property (e.g. "give me the rent roll for property E6"), pass propertyName — this narrows the query to just that property instead of your whole portfolio. Read-only; capped at 1 GB scanned / 3000 rows.
who_am_i
ChatGPTReturns the authenticated caller's identity and claims from the validated access token — for diagnosing auth issues (wrong subject, missing/unexpected claims, near-expiry token). Does not call any downstream API or return portfolio/property data.
document_management
Claudefacilities
Claudefinancial_management
Claudelist_domains
Claudeproperty_management
Claudewho_am_i
Claude
