Using ChatGPT or Claude to Write HaloPSA SQL Reports (copy-paste prompt)

You are writing read-only T-SQL for a HaloPSA custom SQL report (Microsoft SQL Server).

Schema (legacy NetHelpDesk names):
- FAULTS = tickets (also projects, opportunities and tasks). Key columns: Faultid, Symptom (summary), dateoccured (created), datecleared (resolved), Status -> TSTATUS.Tstatus, Areaint -> AREA.Aarea, Assignedtoint -> UNAME.Unum, Requesttype -> REQUESTTYPE.RTid, urgency, impact, fixbydate, FResponseDate, FRespondByDate, Flastactiondate, FDeleted, FexcludefromSLA, InvoiceDate
- ACTIONS = notes, emails and time entries. Join on Faultid. timetaken, ActionChargeHours, ActionNonChargeHours are decimal hours. Charge value is ActionChargeAmount (NOT AChargeTotalAction, which is a bit flag)
- AREA = clients (aareadesc = name). UNAME = agents (uname = name). SITE = sites. USERS = end users
- REQUESTTYPE has RTIsOpportunity and RTIsProject flags

Rules Halo enforces:
1. Always start with SELECT TOP n (ORDER BY fails otherwise, Halo wraps the query as a subquery)
2. No trailing semicolon
3. No -- comments. Use /* */ or none
4. Exclude soft-deleted rows: ISNULL(FDeleted,0) = 0
5. For ticket-only reports join REQUESTTYPE and filter ISNULL(RTIsOpportunity,0) = 0 AND ISNULL(RTIsProject,0) = 0
6. There is no priority column and no SLA breach flag. Breach = datecleared > fixbydate
7. Prefer derived-table JOINs over correlated subqueries in the SELECT list

If you are unsure a column exists, ask me to run:
SELECT TOP 10000 TABLE_NAME, COLUMN_NAME FROM INFORMATION_SCHEMA.COLUMNS WHERE TABLE_NAME = 'FAULTS'

Now write this report: [describe what you want]