ワンクリックで
test-blueprint
Create compact, reviewable high-level designs for test suites before implementation or structural review.
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
Create compact, reviewable high-level designs for test suites before implementation or structural review.
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
Implement provided test blueprints following the project guidelines.
Rules for structuring and connecting Decompose components - regular components, routers, Outputs, factories, navigation, and embedded children
Compose Multiplatform UI patterns for this project - Scaffold screens, AppToolbar, decomposition, CustomTheme, LCE widgets, insets, system bars, messages, core widgets, and previews.
Guidelines for organizing feature module structure using package-by-feature principle
Guidelines for API/DTO boundaries, repository-owned Replica data loading, loaded data observation in components, LCE state, data mutations, cache invalidation, ReplicaAction, ReplicaTag
Compact guide for using the project StringDesc wrapper over Compose Resources
| name | test-blueprint |
| description | Create compact, reviewable high-level designs for test suites before implementation or structural review. |
A Test Blueprint is a compact, human-reviewable part of the implementation plan for a test suite. It describes the SUT name, the test suite type, test names, table cases, and scenario steps before writing executable tests.
A Test Implementation is the executable test code created from the blueprint.
This skill produces blueprint text only. Do not implement tests, edit fixtures, change production code, or run validation as part of this skill.
When the user is planning test work, include the complete blueprint in the final <proposed_plan>.
The blueprint is the test contract that test-implementation follows later during execution.
Use component for Decompose component behavior: screen state, outputs, data loading through DI and mock network, router navigation, root navigation, parameter passing, messages, and dialogs.
Use unit for non-trivial domain logic: algorithms, calculations, validation rules, state reducers, branching/combinatorics.
Use this shape:
<SUT>Test (component)
- observable behavior name
🛠️ Prepare scenario data or test subject
▶️ Perform scenario action or wait for progress
✅ Verify observable result
Rules:
<SUT>Test (component) or <SUT>Test (unit).should in test names.🛠️, ▶️, or ✅.[table] after the test name and followed by a compact cases line:
cases: Pikachu, Charizard, Mewtwo.🛠️ means setup: scenario data, input data, initial state, test subject creation,
capturers, or fakes.▶️ means action: user actions, domain events, navigation requests, lifecycle changes,
waiting for async work, or function behavior under test.✅ means verification: observable state, output events, navigation stack, error state/message,
final calculated result, preserved previous data, or observable external calls.▶️ step, for example: ▶️ Wait for the initial loading to complete.🛠️ steps. Name scenario conditions, not exact fixture names, JSON files, matcher details,
mock server mechanics, or child factory implementation details.▶️ Select an item from the list, not ▶️ Call onItemClick.
Use ▶️ Refresh the list, not ▶️ Call onRefresh.
Use ▶️ Submit the form, not ▶️ Call onSubmitClick.🛠️ Prepare a loaded list, 🛠️ Prepare a form with valid input, or
🛠️ Prepare failed initial loading, not 🛠️ Create RealXxxComponent or exact helper names.✅ Verify the details screen is requested for the selected item, not
✅ Assert output is DetailsRequested. Use ✅ Verify the loaded list is shown, not
✅ Assert state.value.data equals expectedItems.For routers:
handleBackButton, simple pop,
or duplicate safety.Load the closest reference when it helps: