ワンクリックで
honest-degraded-analysis-states
Degrade financial flows visibly: API first, local preview only when warranted, and no invented scenarios
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Degrade financial flows visibly: API first, local preview only when warranted, and no invented scenarios
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Pilot a community Spec Kit extension without pretending it is trusted, official, or a hard gate
Lock what an analytical financial redesign may beautify, but never hide, reorder, or distort
Test tripwires for mortgage analysis UI redesign — preserves logic integrity through visual changes
Choose Spec Kit extensions that reinforce real repo gaps without duplicating existing orchestration
Run Spec Kit through the repo-pinned official CLI and keep the official git extension as the branch workflow source of truth
Freeze cross-cutting contracts before parallel implementation starts
| name | honest-degraded-analysis-states |
| description | Degrade financial flows visibly: API first, local preview only when warranted, and no invented scenarios |
| domain | frontend-integration |
| confidence | high |
| source | earned (Lambert API-first fallback strategy for mortgage comparator MVP) |
Use this skill when the frontend depends on partial backend delivery. In financial UI, degraded states must stay truthful: users should always know whether they are seeing a backend result, a local preview, an unavailable step, or a step that was never requested.
source, label, and detail.api, local_preview, unavailable, and not_requested.apps/web/src/services/analysis-api.ts: API-first orchestration, stage status creation, safe fallback rules, and explicit local-preview notices.apps/web/src/features/mortgage-analysis/analysis-form.tsx: retention guardrails, explicit bonus-variant toggle, and validation that blocks fake bonus scenarios.apps/web/src/features/mortgage-analysis/scenario-comparison-table.tsx: visible two-scenario note instead of an empty or implied bonus row.apps/web/src/services/analysis-api.test.ts: tests prove local fallback is visible, no-bonus stays at two scenarios, and unexpected compare failures do not fabricate output.