[MPDX-9522] Use new staff goal calculations from API

[canac]

Description

Use server-side calculations in the admin view and the presenting your goal step.

Jira ticket: MPDX-9522
Depends on https://github.com/CruGlobal/mpdx_api/pull/3422 deployed 🚀

Testing

• Test 1
  • Go to the coaching page of a new staff goal calculation
  • Check that the 403b amounts are calculated from the percentages
  • Change the percentages and save
  • Check that the 403b amounts change
• Test 2
  • Open a new staff goal calculations
  • Go to step 2
  • Check that the table populates with real data

Checklist:

• I have given my PR a title with the format "MPDX-(JIRA#) (summary sentence max 80 chars)"
• I have applied the appropriate labels (Add the label "Preview" to automatically create a preview environment)
• I have run the Claude Code /pr-review command locally and fixed any relevant suggestions
• I have requested a review from another person on the project
• I have tested my changes in preview or in staging
• I have cleaned up my commit history

[github-actions]

Preview branch generated at https://9522-use-calculations.d3dytjb8adxkk5.amplifyapp.com

[github-actions]

Bundle sizes [mpdx-react]

Compared against 750278f

Route	Size (gzipped)	Diff
/accountLists/[accountListId]/hrTools/nsGoalCalculator	226.88 KB	+1005 B

[canac]

🤖 Multi-Agent Code Review — PR #1873

Verdict: APPROVED WITH SUGGESTIONS — no hard blockers, but one confirmed financial-display divergence should be resolved or product-confirmed before merge. Human review recommended.

6 specialized agents (Financial, Data Integrity, Architecture, Testing, Standards, UX) reviewed in parallel. Testing agent ran the suite: 45/45 pass, tsc exit 0.

Headline

The "Presenting Your Goal" total re-sums a subset of fields client-side and omits the always-applied ×1.06 attrition factor, so the donor-facing "Total Support Goal" systematically understates the admin's authoritative monthlyGoal. See the inline comment on PresentingYourGoalStep.tsx. The established sibling GoalCalculator deliberately folds attrition in (…/PresentingYourGoal.tsx:41 ministryExpenses: ministryExpensesTotal + attrition); this new step drops it and doesn't even query attrition. Tests miss it because the mock is rigged so the 6 fields sum to exactly monthlyGoal (15860).

Verified clean / refuted (no action)

• Null calculations — schema is NewStaffGoalCalculationCalculations! with all Float! subfields; dropping the old ?? 0 guard is correct. Parent is nullable but guarded in both consumers.
• Coaching vs hrTools layout — coaching page renders GoalSettingsForm directly (intended split); hrTools wraps NsGoalCalculator in exactly one NsGoalCalculatorLayout (no double/lost wrap).
• Apollo cache / codegen / mutation refresh — id present, embedded value-object needs none, all 9 fields exist in schema.graphql, mutation re-selects the fragment. Clean.
• Standards — full checklist PASS, no violations. All new strings localized; named exports; no any; TODOs ticket-referenced.

Risk

Mechanical score 10/10 (CRITICAL) is breadth-driven — 12 small sibling files each +1, +2 for the .graphql. Substance is display-only (no auth, no mutations, non-null types), so effective operational risk ≈ MEDIUM, elevated to HIGH by the confirmed financial finding. Auto-approve intentionally does not apply.

Lower-priority items (posted inline)

Fake $3,624 special-needs placeholder shown as real money · 403(b) values missing per-person aria-label · unused goalCalculation prop on ReviewYourCalculationStep (4-agent consensus) · loading/error states render flush to edge · inline error Alert + global snackbar double-surface (informational, not a violation).

Minor (not posted inline)

• Low-value useMemo for monthlyNeeds (PresentingYourGoalStep.tsx:63-74) — consumers don't benefit from a stable ref; optional.
• Hardcoded 'USD' in the 403(b) format — reasonable US-only assumption, consistent with siblings.
• Non-null ! in NsGoalCalculatorTestWrapper.tsx:56 — safe test infra.
• afterEach→afterAll in PresentingYourGoalStep.test.tsx is a correct fix; only residual is the window.print spy no longer restored between tests (harmless today).

_{Generated by /quality:agent-review · standard mode · this is a COMMENT review, not an approval.}

[canac]

🤖 Multi-Agent Code Review — Verdict: APPROVED WITH SUGGESTIONS

6 specialized agents (Architecture, Testing, Standards, Data Integrity, UX, Financial Reporting) + dependency-impact analysis, standard mode. Solid PR: removes mock data, adds real loading/error/empty states, improves type safety, and hoists a remounting layout. Net debt reduced.

No hard blockers. One Important (7.5) finding is strongly recommended before merge (see the inline comment on PresentingYourGoalStep.tsx).

⚠️ 1 Important finding (severity 7.0–7.9) cannot be dismissed via /dismiss. Fix it in this PR or open a follow-up ticket. Medium/Suggestion findings (<7) can be dismissed by replying /dismiss: <reason> to the inline comment.

Risk

Rubric score 10/10 (CRITICAL) — but capped purely by breadth (17 files, 11 medium-risk components) of one cohesive feature. No auth / mutations / cache-policy / server code touched. Practical risk is MEDIUM-HIGH, concentrated in financial-display correctness.

Top finding (Important, 7.5) — agreed by 3 agents + verified against schema

The presentation's "Total Support Goal" is re-derived client-side (useMonthlyNeedsRows rows.reduce) instead of using the authoritative server value calculations.monthlyGoal — which this same PR switched to in GoalSettingsForm.tsx:123. The 6 displayed buckets omit line items 8/10/11/12/13 (staffConferenceTransfer, medicalExpenses, accountTransfers, advocacyTransfers, otherExpenses). Verified from schema.graphql: totalMinistryExpenses = "sum of lines 2-7", monthlyGoal = subtotal(1-13) + adminCharge + attrition. So monthlyGoal − displayedTotal = those five omitted lines. When any is non-zero (medical/other are not senior-only), the staffer is shown a goal lower than reality, and the presentation disagrees with the Goal Settings header. The fixture reconciles only because all five are 0.

Findings summary

#	Sev	Tier	Location
1	7.5	Important	PresentingYourGoalStep.tsx — client reduce vs server monthlyGoal
2	6.0	Medium	PresentingYourGoalStep.test.tsx — afterEach→afterAll isolation regression
3	5.5	Medium	FinancialInformationSection.tsx — aria-label on <p> won't announce
4	5.0	Medium	PresentingYourGoalStep.tsx — attrition folded into Ministry Expenses row
5	4.5	Suggestion	Test: assert total == monthlyGoal (would catch #1)
6	4.0	Suggestion	NsGoalCalculator.tsx — duplicated loading/error/empty block
7	4.0	Suggestion	Test: single-person 403(b) hide path
8	3.5	Suggestion	ReviewYourCalculationStep.tsx — dead goalCalculation prop
9	3.0	Suggestion	NsGoalCalculatorTestWrapper.tsx — non-null assertion (test-only)

Additional low-severity suggestions (not line-anchored): loading Skeleton height={400} doesn't approximate content (kept identical to sibling — consistency nit); support-raised query now waterfalls after goal-calc load on the Presenting step (minor latency).

Dependency impact

No unhandled breaking changes — every consumer updated in-PR or compatible. calculatedResults removed with 0 remaining references repo-wide. Two operational notes: yarn gql codegen must succeed for the fragment change; the default test mock flipped spouseJoining false→true and added fields — run the 4 wrapper-dependent test files (NextStepsStep, NsGoalCalculatorContext, NsGoalCalculatorStepsList, NsGoalCalculatorLayout) to confirm no assertion drift.

Standards

All checklist items PASS (2 low-severity WARN). Named exports ✓ · i18n {{interpolation}} ✓ · <Trans t={t}> ✓ · no any/@ts-ignore/new Date() ✓ · GraphQL id correctly present (calculations is a non-normalizable value object — no id needed) · codegen fields verified present in schema ✓.

[zweatshirt]

Looks good from what I can tell. I only had one real concern. Since this PR will effect #1884 I'll wait for your PR to merge