Skip to content
This repository was archived by the owner on Dec 23, 2025. It is now read-only.

AGENTS.md

Latest commit

 

History

History
143 lines (107 loc) · 4.27 KB

AGENTS.md

File metadata and controls

143 lines (107 loc) · 4.27 KB

AGENTS.md (Context agent guidance)

This file defines specialized “agent hats” you can put on ChatGPT (or multiple GPTs) to work on the app. Each agent has a mission, inputs, outputs, and a prompt template. All agents must respect the domain rules above.

Global guardrails

  • Never break the uniqueness of Assessment(PropertyId, Year).
  • No cascade deletes; use explicit flags/guards.
  • Rerunnable (idempotent) operations wherever possible.
  • Use transactions for Payment + Allocations + Recalc.
  • Prefer small PR‑sized changes; include migration + rollback notes.

Agents

1) Schema Guardian

Purpose: Keep EF entities + migrations aligned with policy. Inputs: Proposed change or new policy. Outputs: *.cs entity diffs, Add-Migration name, and migration code. Template:

You are Schema Guardian. Given a change, produce:
1) Entity class edits
2) Fluent config diffs (OnModelCreating)
3) EF migration (Up/Down)
4) Data backfill script if needed

2) Rollover Captain

Purpose: Year rollover (create missing assessments, set fees, recalc). Inputs: Target year. Outputs: Service method, LINQ/SQL, page flow. Template:

Design an idempotent RolloverService.CreateMissingAssessments(year) that:
- Ensures FeeSchedule[year]
- Creates Assessments only for active/in‑jurisdiction properties
- Leaves existing rows untouched
- Calls RecalcService.Recompute(year)
Provide tests.

3) Allocation Wizard

Purpose: Allocate Payments to open Assessments (and donations). Inputs: Payment DTO (person, date, amount, type, memo). Outputs: Allocation algorithm, edge‑case handling. Template:

Implement PaymentService.Allocate(paymentId, policy) with:
- Current‑year default targeting
- Exact cover fast‑path
- Partial/overpay behavior
- Transaction + concurrency notes
- Unit tests for split allocations

4) Utility Clerk

Purpose: Import utility CSV, match aliases, auto‑allocate exact cover. Inputs: CSV sample; fee for year. Outputs: Parser, matcher (score), auto vs review workflow. Template:

Create UtilityService.Import(csv) -> List<Notice>
Create UtilityService.SuggestMatches(noticeId) -> (PersonId, score)
Auto‑allocate when Amount == OwnedCount × Fee; else flag NeedsReview

5) Mailroom Clerk

Purpose: Handle returns, promote next valid primary. Inputs: PersonId, AddressId. Outputs: Update logic, UI actions, test. Template:

AddressService.MarkReturned(addressId):
- Set IsValidForMail=false
- If Primary -> promote next IsValidForMail=true
- If none -> person appears on NoMailingAddress report

6) Report Builder

Purpose: Build on‑screen grids + PDF/labels. Inputs: Report type + filters. Outputs: LINQ/SQL, QuestPDF/Razor page, export method. Template:

Produce OwnersWhoOwe(year): Person, Properties[], AmountDue, AmountPaid, Balance
Export to PDF and Avery labels (5160/8160)

7) QA Tester

Purpose: Fixtures + tests for services. Inputs: Target service. Outputs: xUnit tests, seed data, edge cases. Template:

Given PaymentService, author tests for: exact cover, underpay, overpay, donation, multiple properties, concurrency.

Prompt snippets

  • Use DB:
Assume SQLite + EF Core. Provide migration code and a reversible Down(). Use Restrict deletes. Create unique index on Assessment(PropertyId, Year).
  • Use Blazor:
Provide a Razor component with EditForm, DataAnnotations validation, and async service calls. Include minimal styling and error handling.
  • Backup pattern:
Show a method that runs `VACUUM INTO` to a timestamped path, with try/catch and a success toast.

Definition of Done (per PR)

  • Compiles; migrations apply cleanly on a fresh DB.
  • Unit tests pass; at least one PDF/labels export verified.
  • Rerunning the action causes no duplicates or data corruption.
  • Logging and basic error messages in place.

Build Notes

  • All dotnet build tasks need to run with escalated priviledges.
  • Do not create migrations in code by hand. You should prompt the user to run dotnet ef migrations add instead.
  • You are encouraged to run dotnet build after making code changes to verify it was built correctly. If there are errors, attempt to fix them before completing the task.