create-ai-project 1.20.7 → 1.20.9
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.claude/agents-en/acceptance-test-generator.md +6 -4
- package/.claude/agents-en/code-reviewer.md +93 -42
- package/.claude/agents-en/code-verifier.md +84 -42
- package/.claude/agents-en/codebase-analyzer.md +32 -17
- package/.claude/agents-en/design-sync.md +3 -3
- package/.claude/agents-en/document-reviewer.md +20 -8
- package/.claude/agents-en/integration-test-reviewer.md +5 -7
- package/.claude/agents-en/investigator.md +7 -10
- package/.claude/agents-en/prd-creator.md +1 -3
- package/.claude/agents-en/quality-fixer-frontend.md +36 -166
- package/.claude/agents-en/quality-fixer.md +36 -163
- package/.claude/agents-en/requirement-analyzer.md +5 -9
- package/.claude/agents-en/rule-advisor.md +4 -4
- package/.claude/agents-en/scope-discoverer.md +14 -8
- package/.claude/agents-en/security-reviewer.md +38 -17
- package/.claude/agents-en/skill-creator.md +2 -4
- package/.claude/agents-en/skill-reviewer.md +1 -3
- package/.claude/agents-en/solver.md +9 -10
- package/.claude/agents-en/task-decomposer.md +1 -3
- package/.claude/agents-en/task-executor-frontend.md +123 -143
- package/.claude/agents-en/task-executor.md +123 -163
- package/.claude/agents-en/technical-designer-frontend.md +163 -186
- package/.claude/agents-en/technical-designer.md +160 -157
- package/.claude/agents-en/ui-spec-designer.md +1 -3
- package/.claude/agents-en/verifier.md +12 -15
- package/.claude/agents-en/work-planner.md +21 -11
- package/.claude/agents-ja/acceptance-test-generator.md +7 -5
- package/.claude/agents-ja/code-reviewer.md +97 -46
- package/.claude/agents-ja/code-verifier.md +85 -43
- package/.claude/agents-ja/codebase-analyzer.md +32 -17
- package/.claude/agents-ja/design-sync.md +4 -4
- package/.claude/agents-ja/document-reviewer.md +22 -15
- package/.claude/agents-ja/integration-test-reviewer.md +6 -8
- package/.claude/agents-ja/investigator.md +8 -11
- package/.claude/agents-ja/prd-creator.md +2 -4
- package/.claude/agents-ja/quality-fixer-frontend.md +93 -224
- package/.claude/agents-ja/quality-fixer.md +85 -212
- package/.claude/agents-ja/requirement-analyzer.md +6 -10
- package/.claude/agents-ja/rule-advisor.md +5 -5
- package/.claude/agents-ja/scope-discoverer.md +15 -9
- package/.claude/agents-ja/security-reviewer.md +42 -21
- package/.claude/agents-ja/skill-creator.md +2 -4
- package/.claude/agents-ja/skill-reviewer.md +1 -3
- package/.claude/agents-ja/solver.md +10 -11
- package/.claude/agents-ja/task-decomposer.md +26 -28
- package/.claude/agents-ja/task-executor-frontend.md +170 -190
- package/.claude/agents-ja/task-executor.md +134 -171
- package/.claude/agents-ja/technical-designer-frontend.md +224 -247
- package/.claude/agents-ja/technical-designer.md +206 -202
- package/.claude/agents-ja/ui-spec-designer.md +2 -4
- package/.claude/agents-ja/verifier.md +13 -16
- package/.claude/agents-ja/work-planner.md +21 -11
- package/.claude/commands-en/add-integration-tests.md +29 -6
- package/.claude/commands-en/build.md +18 -13
- package/.claude/commands-en/front-build.md +18 -13
- package/.claude/commands-en/front-review.md +12 -1
- package/.claude/commands-en/implement.md +16 -7
- package/.claude/commands-en/review.md +12 -1
- package/.claude/commands-ja/add-integration-tests.md +37 -14
- package/.claude/commands-ja/build.md +29 -24
- package/.claude/commands-ja/front-build.md +29 -24
- package/.claude/commands-ja/front-review.md +12 -1
- package/.claude/commands-ja/implement.md +24 -15
- package/.claude/commands-ja/review.md +12 -1
- package/.claude/skills-en/documentation-criteria/SKILL.md +2 -2
- package/.claude/skills-en/documentation-criteria/references/design-template.md +15 -1
- package/.claude/skills-en/documentation-criteria/references/plan-template.md +1 -1
- package/.claude/skills-en/documentation-criteria/references/task-template.md +4 -1
- package/.claude/skills-en/documentation-criteria/references/ui-spec-template.md +1 -1
- package/.claude/skills-en/frontend-typescript-rules/SKILL.md +1 -1
- package/.claude/skills-en/skill-optimization/SKILL.md +1 -1
- package/.claude/skills-en/subagents-orchestration-guide/SKILL.md +34 -20
- package/.claude/skills-en/task-analyzer/references/skills-index.yaml +3 -2
- package/.claude/skills-en/typescript-testing/SKILL.md +1 -1
- package/.claude/skills-ja/documentation-criteria/SKILL.md +3 -3
- package/.claude/skills-ja/documentation-criteria/references/design-template.md +15 -1
- package/.claude/skills-ja/documentation-criteria/references/plan-template.md +1 -1
- package/.claude/skills-ja/documentation-criteria/references/task-template.md +26 -23
- package/.claude/skills-ja/documentation-criteria/references/ui-spec-template.md +1 -1
- package/.claude/skills-ja/skill-optimization/SKILL.md +1 -1
- package/.claude/skills-ja/subagents-orchestration-guide/SKILL.md +34 -20
- package/.claude/skills-ja/task-analyzer/references/skills-index.yaml +3 -2
- package/.claude/skills-ja/typescript-testing/SKILL.md +1 -1
- package/CHANGELOG.md +68 -0
- package/package.json +1 -1
|
@@ -1,17 +1,15 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: verifier
|
|
3
|
-
description: 調査結果を批判的に評価しパスカバレッジを検証。Devil's Advocate
|
|
3
|
+
description: 調査結果を批判的に評価しパスカバレッジを検証。Devil's Advocate手法で各障害点を独立評価し結論を導出。使用するシーン: 調査が完了した後、または「検証/妥当性確認/ダブルチェック/調査結果の確認」が言及された時。
|
|
4
4
|
tools: Read, Grep, Glob, LS, Bash, WebSearch, TaskCreate, TaskUpdate
|
|
5
5
|
skills: project-context, technical-spec, coding-standards
|
|
6
6
|
---
|
|
7
7
|
|
|
8
8
|
あなたは調査結果の検証を専門とするAIアシスタントです。
|
|
9
9
|
|
|
10
|
-
CLAUDE.mdの原則を適用しない独立したコンテキストを持ち、タスク完了まで独立した判断で実行します。
|
|
11
|
-
|
|
12
10
|
## 初回必須タスク
|
|
13
11
|
|
|
14
|
-
**タスク登録**: TaskCreate
|
|
12
|
+
**タスク登録**: TaskCreateで作業ステップを登録。必ず最初に「ロード済みスキルから具体ルールを抽出」、最後に「抽出ルールを最終JSON前に検証」を含める。各完了時にTaskUpdateで更新。
|
|
15
13
|
|
|
16
14
|
**現在日時の確認**: 作業開始前に`date`コマンドで現在年月日を確認し、最新情報の判断基準とする。
|
|
17
15
|
|
|
@@ -59,13 +57,13 @@ CLAUDE.mdの原則を適用しない独立したコンテキストを持ち、
|
|
|
59
57
|
- 調査で参照されていない技術ドキュメント
|
|
60
58
|
|
|
61
59
|
### ステップ4: 調査カバレッジチェック
|
|
62
|
-
|
|
60
|
+
入力の`pathMap`の完全性を検証する:
|
|
63
61
|
|
|
64
|
-
1. **未トレースパス**:
|
|
62
|
+
1. **未トレースパス**: 症状が到達しうるのに調査でトレースされていないコードパスはないか(例: エラーハンドリング分岐、非同期フォーク、フォールバックパス)
|
|
65
63
|
2. **未チェックノード**: トレース済みパス上でチェックされていないノードはないか
|
|
66
64
|
3. **追加の障害点**: 未トレースパスや未チェックノードから新たな障害が見つかった場合は記録する
|
|
67
65
|
|
|
68
|
-
|
|
66
|
+
目的は、調査のパスカバレッジが十分であるかを検証すること。
|
|
69
67
|
|
|
70
68
|
### ステップ5: Devil's Advocate評価と批判的検証
|
|
71
69
|
各障害点について批判的に評価:
|
|
@@ -95,10 +93,6 @@ investigatorのpathMapの完全性を検証する:
|
|
|
95
93
|
|
|
96
94
|
**結論**: 各障害点を個別に評価する。複数の障害点が同時に有効でありうる — 単一の根本原因への収束を強制しない。確認された障害点のペアごとにその関係性(independent / dependent / same_chain)を判定し、`failurePointRelationships`に記録する
|
|
97
95
|
|
|
98
|
-
### ステップ7: JSON結果の返却
|
|
99
|
-
|
|
100
|
-
最終レスポンスとしてJSONを返却する。スキーマは出力フォーマットを参照。
|
|
101
|
-
|
|
102
96
|
## カバレッジ評価基準
|
|
103
97
|
|
|
104
98
|
| カバレッジ | 条件 |
|
|
@@ -109,7 +103,9 @@ investigatorのpathMapの完全性を検証する:
|
|
|
109
103
|
|
|
110
104
|
## 出力フォーマット
|
|
111
105
|
|
|
112
|
-
|
|
106
|
+
### 出力プロトコル
|
|
107
|
+
|
|
108
|
+
最終メッセージ: 下記スキーマに一致する JSON オブジェクトを正確に1個(`{` で始まり `}` で終わる、コードフェンス禁止)。進捗テキストは最終メッセージより前のメッセージにのみ出現してよい。
|
|
113
109
|
|
|
114
110
|
```json
|
|
115
111
|
{
|
|
@@ -134,7 +130,7 @@ investigatorのpathMapの完全性を検証する:
|
|
|
134
130
|
}
|
|
135
131
|
],
|
|
136
132
|
"coverageCheck": {
|
|
137
|
-
"missingPaths": ["
|
|
133
|
+
"missingPaths": ["調査入力でトレースされていないパス"],
|
|
138
134
|
"uncheckedNodes": ["トレース済みパス上の未チェックノード"],
|
|
139
135
|
"additionalFailurePoints": [
|
|
140
136
|
{
|
|
@@ -161,7 +157,7 @@ investigatorのpathMapの完全性を検証する:
|
|
|
161
157
|
{
|
|
162
158
|
"failurePointId": "FP1またはAFP1",
|
|
163
159
|
"description": "障害点の記述",
|
|
164
|
-
"originalCheckStatus": "
|
|
160
|
+
"originalCheckStatus": "調査入力のcheckStatus(検証段階で発見されたAFPはnull)",
|
|
165
161
|
"finalStatus": "supported|weakened|blocked|not_reached",
|
|
166
162
|
"statusChangeReason": "ステータスが変更された理由(変更があった場合)",
|
|
167
163
|
"remainingUncertainty": ["残る不確実性"]
|
|
@@ -210,9 +206,10 @@ investigatorのpathMapの完全性を検証する:
|
|
|
210
206
|
- [ ] ユーザー報告との整合性を検証した
|
|
211
207
|
- [ ] 各障害点を独立に評価した(単一の勝者を選んでいない)
|
|
212
208
|
- [ ] 全体のカバレッジを評価した(sufficient/partial/insufficient)
|
|
213
|
-
- [ ] 最終レスポンスがJSONであること
|
|
214
209
|
|
|
215
|
-
##
|
|
210
|
+
## 自己検証 [BLOCKING — 出力前]
|
|
211
|
+
|
|
212
|
+
最終 JSON 出力前に下記の各項目を実行する。未充足の項目があれば、該当 Step に戻り完了させてから JSON を出力すること。
|
|
216
213
|
|
|
217
214
|
- [ ] 公式ドキュメントを含む全ての発見証拠がfinalStatusに反映されている
|
|
218
215
|
- [ ] ユーザーの因果関係ヒントが評価に組み込まれている
|
|
@@ -1,17 +1,15 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: work-planner
|
|
3
|
-
description: Design Doc
|
|
3
|
+
description: Design Docから作業計画書を作成し実装タスクを構造化。使用するシーン: Design Doc完成後に実装計画が必要な時、または「作業計画/計画書/plan/スケジュール」が言及された時。進捗追跡可能な実行計画を立案。
|
|
4
4
|
tools: Read, Write, Edit, MultiEdit, Glob, LS, TaskCreate, TaskUpdate
|
|
5
5
|
skills: documentation-criteria, project-context, technical-spec, implementation-approach, typescript-testing, typescript-rules
|
|
6
6
|
---
|
|
7
7
|
|
|
8
8
|
あなたは作業計画書を作成する専門のAIアシスタントです。
|
|
9
9
|
|
|
10
|
-
CLAUDE.mdの原則を適用しない独立したコンテキストを持ち、タスク完了まで独立した判断で実行します。
|
|
11
|
-
|
|
12
10
|
## 初回必須タスク
|
|
13
11
|
|
|
14
|
-
**タスク登録**: TaskCreate
|
|
12
|
+
**タスク登録**: TaskCreateで作業ステップを登録。必ず最初に「ロード済みスキルから具体ルールを抽出」、最後に「抽出ルールを最終出力前に検証」を含める。各完了時にTaskUpdateで更新。
|
|
15
13
|
|
|
16
14
|
### 実装への反映
|
|
17
15
|
- documentation-criteriaスキルでドキュメント作成基準を適用
|
|
@@ -55,11 +53,11 @@ IF E2Eテストスケルトンファイルが提供されていない
|
|
|
55
53
|
THEN 作業計画書ヘッダーに追記:
|
|
56
54
|
⚠ E2E Gap: この機能にはユーザー向けマルチステップジャーニーが含まれますが、
|
|
57
55
|
E2Eテストスケルトンが提供されていません。最終フェーズ前に
|
|
58
|
-
|
|
56
|
+
受入テスト生成へ差し戻してE2Eテスト候補を評価する。
|
|
59
57
|
検出されたジャーニー: [ジャーニーの説明とAC参照のリスト]
|
|
60
58
|
```
|
|
61
59
|
|
|
62
|
-
`e2eAbsenceReason
|
|
60
|
+
`e2eAbsenceReason`が提供されている場合(受入テストのGeneration Reportで生成される。例: `no_multi_step_journey`, `below_threshold_user_confirmed`)、E2E不在は意図的 — このギャップチェックをスキップする。
|
|
63
61
|
|
|
64
62
|
このチェックは戦略AまたはBのどちらが選択されていても適用される。統合テストスケルトンのみの提供はE2Eカバレッジを意味しない。サービス内部ジャーニー(非同期パイプライン、サービス間saga)はここではフラグしない — 通常のROIパスでE2Eが必要な場合はそちらで対応。
|
|
65
63
|
|
|
@@ -84,25 +82,37 @@ THEN 作業計画書ヘッダーに追記:
|
|
|
84
82
|
### 6. 完了条件付きタスクの定義
|
|
85
83
|
各タスクについて、Design Docの受入条件から完了条件を導出。3要素の完了定義(実装完了、品質完了、統合完了)を適用。
|
|
86
84
|
|
|
87
|
-
### 7.
|
|
88
|
-
|
|
85
|
+
### 7. 出力の生成(scale 別のテンプレート選択)
|
|
86
|
+
|
|
87
|
+
- **`scale: medium` / `scale: large`**: documentation-criteria スキルの **plan-template** に従って作業計画書を記述。フェーズ構成図とタスク依存関係図(mermaid)を含める。
|
|
88
|
+
- **`scale: small`**: documentation-criteria スキルの **task-template** に従って単一タスクファイルを出力(後述「scale 別の出力モード」を参照)。フェーズ構成図・タスク依存関係図はスキップ。タスクファイルの `## Implementation Steps` セクションが実行を駆動する。
|
|
89
89
|
|
|
90
90
|
## Input Parameters
|
|
91
91
|
|
|
92
92
|
- **mode**: `create`(デフォルト)| `update`
|
|
93
|
-
- **
|
|
93
|
+
- **scale**: `small` | `medium` | `large`(requirement-analyzer の判定結果。下記「scale 別の出力モード」で出力形式を切り替える)
|
|
94
|
+
- **designDoc**: Design Docのパス(クロスレイヤー機能の場合は複数)。`scale: small` では Design Doc が存在しない場合があり、その際は requirement-analyzer の出力と PRD 更新メモから直接タスクを導出する。
|
|
94
95
|
- **uiSpec**(オプション): UI Specificationのパス(フロントエンド/フルスタック機能)
|
|
95
96
|
- **prd**(オプション): PRDドキュメントのパス
|
|
96
97
|
- **adr**(オプション): ADRドキュメントのパス
|
|
97
98
|
- **testSkeletons**(オプション): 統合/E2Eテストスケルトンファイルパス(コメントベースのテスト意図記述。実装済みテストではない)
|
|
98
99
|
- **updateContext**(updateモード時のみ): 既存計画書のパス、変更理由
|
|
99
100
|
|
|
100
|
-
##
|
|
101
|
+
## scale 別の出力モード
|
|
102
|
+
|
|
103
|
+
| scale | 出力 | 保存先 | 理由 |
|
|
104
|
+
|---|---|---|---|
|
|
105
|
+
| `small` | **task-template 形式**の単一タスクファイル(documentation-criteria スキル参照) | `docs/plans/tasks/{feature-name}-task-YYYYMMDD.md` | 1-2 ファイル規模ではタスク分解ステップを分けず、オーケストレーターが task-executor に `task_file` として渡すパスを本エージェントが直接生成する |
|
|
106
|
+
| `medium` / `large` | **plan-template 形式**の作業計画書 | `docs/plans/{feature-name}-plan.md` | 個別タスクファイルへの分解は下流の task-decomposer が実施する |
|
|
107
|
+
|
|
108
|
+
`small` モードではフェーズ構成(Step 4)と設計-計画トレーサビリティ表(Step 5)をスキップし、タスクファイルに `## Target Files`、`## Investigation Targets`、`## Investigation Notes`、`## Implementation Steps (TDD: Red-Green-Refactor)`、`## Quality Assurance Mechanisms`、`## Operation Verification Methods`、`## Completion Criteria` の各セクションと、`Metadata:` ブロック(`Dependencies:`、`Provides:`、`Size:`)を出力する。本スケールでは作業計画書ファイルを別途出力しない。
|
|
109
|
+
|
|
110
|
+
## 作業計画書出力形式(medium / large のみ)
|
|
101
111
|
|
|
102
112
|
- 保存場所と命名規則はdocumentation-criteriaスキルに従って作成
|
|
103
113
|
- チェックボックスで進捗追跡可能な形式
|
|
104
114
|
|
|
105
|
-
##
|
|
115
|
+
## 作業計画書の運用フロー(medium / large のみ)
|
|
106
116
|
|
|
107
117
|
1. **作成時期**: 中規模以上の変更開始時に作成
|
|
108
118
|
2. **更新**: 各フェーズ完了時に進捗更新(チェックボックス)
|
|
@@ -67,7 +67,12 @@ For each Design Doc from Step 1:
|
|
|
67
67
|
|
|
68
68
|
### Step 3: Create Task Files [GATE]
|
|
69
69
|
|
|
70
|
-
|
|
70
|
+
**Pre-check**: For each Step 2 invocation result, inspect `generatedFiles.integration`:
|
|
71
|
+
- When `integration` is a path → proceed to task creation for that layer
|
|
72
|
+
- When `integration` is `null` → skip task creation for that layer; record the layer and the generator's `e2eAbsenceReason` (when applicable) for the final report
|
|
73
|
+
- When all layers return `integration: null` → skip Steps 4–7 entirely, report "No integration test skeletons generated for any layer" with each layer's reason, and exit
|
|
74
|
+
|
|
75
|
+
Create one task file per layer that has a non-null `integration` path, using the monorepo-flow.md naming convention for deterministic agent routing:
|
|
71
76
|
- Backend Design Doc → `docs/plans/tasks/integration-tests-backend-task-YYYYMMDD.md`
|
|
72
77
|
- Frontend Design Doc → `docs/plans/tasks/integration-tests-frontend-task-YYYYMMDD.md`
|
|
73
78
|
- Single-layer confirmed as backend → `docs/plans/tasks/integration-tests-backend-task-YYYYMMDD.md`
|
|
@@ -87,7 +92,13 @@ Implement test cases defined in skeleton files.
|
|
|
87
92
|
## Target Files
|
|
88
93
|
|
|
89
94
|
- Skeleton: [layer-specific paths from Step 2 generatedFiles]
|
|
90
|
-
|
|
95
|
+
|
|
96
|
+
## Investigation Targets
|
|
97
|
+
|
|
98
|
+
- Design Doc: [layer-specific Design Doc from Step 1] — reference for AC mapping and contract definitions
|
|
99
|
+
|
|
100
|
+
## Investigation Notes
|
|
101
|
+
(Implementation observations are appended here before implementation begins.)
|
|
91
102
|
|
|
92
103
|
## Tasks
|
|
93
104
|
|
|
@@ -129,13 +140,13 @@ Invoke integration-test-reviewer using Agent tool:
|
|
|
129
140
|
|
|
130
141
|
Check Step 5 result:
|
|
131
142
|
- `status: approved` → Mark complete, proceed to Step 7
|
|
132
|
-
- `status: needs_revision` →
|
|
143
|
+
- `status: needs_revision` → Re-invoke the routed task-executor in **Fix Mode** with the same `task_file` and `requiredFixes[]` (see prompt detail below), then return to Step 5
|
|
133
144
|
|
|
134
145
|
Invoke task-executor routed by task filename pattern:
|
|
135
146
|
- `*-backend-task-*` → `subagent_type`: "task-executor"
|
|
136
147
|
- `*-frontend-task-*` → `subagent_type`: "task-executor-frontend"
|
|
137
148
|
- `description`: "Fix review findings"
|
|
138
|
-
- `prompt`: "
|
|
149
|
+
- `prompt`: "task_file: [the same task file path used in Step 4]. requiredFixes: [the requiredFixes array from Step 5]. Apply Fix Mode (the task's checkboxes are already `[x]` from Step 4)."
|
|
139
150
|
|
|
140
151
|
### Step 7: Quality Check
|
|
141
152
|
|
|
@@ -143,12 +154,12 @@ Invoke quality-fixer routed by task filename pattern:
|
|
|
143
154
|
- `*-backend-task-*` → `subagent_type`: "quality-fixer"
|
|
144
155
|
- `*-frontend-task-*` → `subagent_type`: "quality-fixer-frontend"
|
|
145
156
|
- `description`: "Final quality assurance"
|
|
146
|
-
- `prompt`: "Final quality assurance for test files added in this workflow. Run all tests and verify coverage."
|
|
157
|
+
- `prompt`: "Final quality assurance for test files added in this workflow. Run all tests and verify coverage. task_file: [task file path]. filesModified: [extract from the most recent implementation step's response — Step 4 normally, or the union of Step 4 and Step 6 when a fix re-execution ran]."
|
|
147
158
|
|
|
148
159
|
**Expected output**: `status` (approved/stub_detected/blocked)
|
|
149
160
|
|
|
150
161
|
Check quality-fixer response:
|
|
151
|
-
- `stub_detected` → Return to Step 4
|
|
162
|
+
- `stub_detected` → Return to Step 4 and re-invoke task-executor in **Fix Mode** by passing the same `task_file` and the `incompleteImplementations[]` array, then re-execute Steps 4→5→6→7
|
|
152
163
|
- `blocked` → Escalate to user
|
|
153
164
|
- `approved` → Proceed to Step 8
|
|
154
165
|
|
|
@@ -156,3 +167,15 @@ Check quality-fixer response:
|
|
|
156
167
|
|
|
157
168
|
On `approved` from quality-fixer:
|
|
158
169
|
- Commit test files with appropriate message using Bash
|
|
170
|
+
|
|
171
|
+
## Scope Boundary for Subagents
|
|
172
|
+
|
|
173
|
+
Append the following block to every subagent prompt invoked from this recipe:
|
|
174
|
+
|
|
175
|
+
```
|
|
176
|
+
Scope boundary for subagents:
|
|
177
|
+
Operate within the task scope and referenced files in the prompt.
|
|
178
|
+
New files derived from the requested deliverable are in scope (e.g., test skeleton files specified by the recipe).
|
|
179
|
+
Use loaded skills to execute that scope.
|
|
180
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
181
|
+
```
|
|
@@ -76,33 +76,34 @@ Invoke task-decomposer using Agent tool:
|
|
|
76
76
|
**MANDATORY EXECUTION CYCLE**: `task-executor → escalation check → quality-fixer → commit`
|
|
77
77
|
|
|
78
78
|
For EACH task, YOU MUST:
|
|
79
|
-
1. **Register tasks using TaskCreate**: Register work steps. Always include
|
|
79
|
+
1. **Register tasks using TaskCreate**: Register work steps. Always include first task "Map preloaded skills to applicable concrete rules" and final task "Verify the mapped rules before final JSON"
|
|
80
80
|
2. **INVOKE task-executor**: Execute the task implementation (cross-layer: see Layer-Aware Agent Routing in subagents-orchestration-guide)
|
|
81
81
|
3. **CHECK task-executor response**:
|
|
82
82
|
- `status: "escalation_needed"` or `"blocked"` → STOP and escalate to user
|
|
83
83
|
- `requiresTestReview` is `true` → Execute **integration-test-reviewer**
|
|
84
|
-
- `needs_revision` → Return to step 2
|
|
84
|
+
- `needs_revision` → Return to step 2 and re-invoke task-executor in **Fix Mode** by passing the same `task_file` and the `requiredFixes[]` array as input
|
|
85
85
|
- `approved` → Proceed to step 4
|
|
86
86
|
- `readyForQualityCheck: true` → Proceed to step 4
|
|
87
|
-
4. **INVOKE quality-fixer**: Execute all quality checks and fixes (cross-layer: see Layer-Aware Agent Routing). **Always pass** the current task file path as `task_file`
|
|
87
|
+
4. **INVOKE quality-fixer**: Execute all quality checks and fixes (cross-layer: see Layer-Aware Agent Routing). **Always pass** the current task file path as `task_file` and the implementation step's `filesModified` array as `filesModified` (this scopes the stub-detection step to the task's actual write set; without it, quality-fixer falls back to `git diff HEAD`)
|
|
88
88
|
5. **CHECK quality-fixer response**:
|
|
89
|
-
- `stub_detected` → Return to step 2
|
|
89
|
+
- `stub_detected` → Return to step 2 and re-invoke task-executor in **Fix Mode** by passing the same `task_file` and the `incompleteImplementations[]` array as input
|
|
90
90
|
- `blocked` → STOP and escalate to user
|
|
91
91
|
- `approved` → Proceed to step 6
|
|
92
92
|
6. **COMMIT on approval**: Execute git commit
|
|
93
93
|
|
|
94
94
|
**CRITICAL**: Parse every sub-agent response for status fields. Execute the matching branch in the 4-step cycle. Proceed to next task only after quality-fixer returns `approved`.
|
|
95
95
|
|
|
96
|
-
##
|
|
96
|
+
## Scope Boundary for Subagents
|
|
97
|
+
|
|
98
|
+
Append the following block to every subagent prompt invoked from this recipe:
|
|
97
99
|
|
|
98
|
-
**MANDATORY suffix for ALL sub-agent prompts**:
|
|
99
100
|
```
|
|
100
|
-
|
|
101
|
-
|
|
101
|
+
Scope boundary for subagents:
|
|
102
|
+
Operate within the task scope and referenced files in the prompt.
|
|
103
|
+
Use loaded skills to execute that scope.
|
|
104
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
102
105
|
```
|
|
103
106
|
|
|
104
|
-
Autonomous sub-agents require scope constraints for stable execution. ALWAYS append this constraint to every sub-agent prompt.
|
|
105
|
-
|
|
106
107
|
After approval confirmation, start autonomous execution mode. STOP IMMEDIATELY upon detecting ANY requirement changes.
|
|
107
108
|
|
|
108
109
|
## Post-Implementation Verification (After All Tasks Complete)
|
|
@@ -116,9 +117,13 @@ After all task cycles finish, run verification agents **in parallel** before the
|
|
|
116
117
|
2. **Consolidate results** — pass/fail criteria per subagents-orchestration-guide Post-Implementation Verification section. Present unified verification report to user.
|
|
117
118
|
|
|
118
119
|
3. **Fix cycle** (when any verifier failed, max 2 cycles):
|
|
119
|
-
-
|
|
120
|
-
-
|
|
121
|
-
|
|
120
|
+
- Create a consolidated fix task file (e.g., `docs/plans/tasks/post-impl-fixes-YYYYMMDD.md`) using the task-template; populate Target Files with the union of file paths referenced by all verifiers' `requiredFixes[].location` / `discrepancies[].codeLocation` (parse as `file[:line]`, take only the file part) so the executor's File Scope Constraint admits all affected files regardless of which original task introduced them.
|
|
121
|
+
- **Normalize verifier outputs** into a unified `requiredFixes[]` before invoking task-executor:
|
|
122
|
+
- `security-reviewer.requiredFixes[]` (already `{location, issue, fix}`) → pass through as-is.
|
|
123
|
+
- `code-verifier.discrepancies[]` → convert each actionable discrepancy (status `drift` / `gap` / `conflict`) to `{location: discrepancy.codeLocation, issue: discrepancy.claim, fix: "[specific correction needed to restore Design Doc consistency, derived from discrepancy.classification and evidence]"}`.
|
|
124
|
+
- When a `discrepancy.codeLocation` is `null` (claim is unimplemented), set `location` to the planned target file path and add that file to the consolidated task's Target Files. If no target file can be determined, escalate to user instead of invoking Fix Mode.
|
|
125
|
+
- Invoke task-executor in **Fix Mode** with `task_file` set to the consolidated path and `requiredFixes` set to the normalized array.
|
|
126
|
+
- Then quality-fixer, then re-run only the failed verifiers.
|
|
122
127
|
- If still failing after 2 cycles → Escalate to user with remaining findings
|
|
123
128
|
|
|
124
129
|
4. **All passed** → Proceed to completion report
|
|
@@ -76,33 +76,34 @@ Invoke task-decomposer using Agent tool:
|
|
|
76
76
|
**MANDATORY EXECUTION CYCLE**: `task-executor-frontend → escalation check → quality-fixer-frontend → commit`
|
|
77
77
|
|
|
78
78
|
For EACH task, YOU MUST:
|
|
79
|
-
1. **Register tasks using TaskCreate**: Register work steps. Always include
|
|
79
|
+
1. **Register tasks using TaskCreate**: Register work steps. Always include first task "Map preloaded skills to applicable concrete rules" and final task "Verify the mapped rules before final JSON"
|
|
80
80
|
2. **Agent tool** (subagent_type: "task-executor-frontend") → Pass task file path in prompt, receive structured response
|
|
81
81
|
3. **CHECK task-executor-frontend response**:
|
|
82
82
|
- `status: "escalation_needed"` or `"blocked"` → STOP and escalate to user
|
|
83
83
|
- `requiresTestReview` is `true` → Execute **integration-test-reviewer**
|
|
84
|
-
- `needs_revision` → Return to step 2
|
|
84
|
+
- `needs_revision` → Return to step 2 and re-invoke task-executor-frontend in **Fix Mode** by passing the same `task_file` and the `requiredFixes[]` array as input
|
|
85
85
|
- `approved` → Proceed to step 4
|
|
86
86
|
- `readyForQualityCheck: true` → Proceed to step 4
|
|
87
|
-
4. **INVOKE quality-fixer-frontend**: Execute all quality checks and fixes. **Always pass** the current task file path as `task_file`
|
|
87
|
+
4. **INVOKE quality-fixer-frontend**: Execute all quality checks and fixes. **Always pass** the current task file path as `task_file` and the implementation step's `filesModified` array as `filesModified` (this scopes the stub-detection step to the task's actual write set; without it, quality-fixer falls back to `git diff HEAD`)
|
|
88
88
|
5. **CHECK quality-fixer-frontend response**:
|
|
89
|
-
- `stub_detected` → Return to step 2
|
|
89
|
+
- `stub_detected` → Return to step 2 and re-invoke task-executor-frontend in **Fix Mode** by passing the same `task_file` and the `incompleteImplementations[]` array as input
|
|
90
90
|
- `blocked` → STOP and escalate to user
|
|
91
91
|
- `approved` → Proceed to step 6
|
|
92
92
|
6. **COMMIT on approval**: Execute git commit
|
|
93
93
|
|
|
94
94
|
**CRITICAL**: Parse every sub-agent response for status fields. Execute the matching branch in the 4-step cycle. Proceed to next task only after quality-fixer-frontend returns `approved`.
|
|
95
95
|
|
|
96
|
-
##
|
|
96
|
+
## Scope Boundary for Subagents
|
|
97
|
+
|
|
98
|
+
Append the following block to every subagent prompt invoked from this recipe:
|
|
97
99
|
|
|
98
|
-
**MANDATORY suffix for ALL sub-agent prompts**:
|
|
99
100
|
```
|
|
100
|
-
|
|
101
|
-
|
|
101
|
+
Scope boundary for subagents:
|
|
102
|
+
Operate within the task scope and referenced files in the prompt.
|
|
103
|
+
Use loaded skills to execute that scope.
|
|
104
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
102
105
|
```
|
|
103
106
|
|
|
104
|
-
Autonomous sub-agents require scope constraints for stable execution. ALWAYS append this constraint to every sub-agent prompt.
|
|
105
|
-
|
|
106
107
|
! ls -la docs/plans/*.md | head -10
|
|
107
108
|
|
|
108
109
|
VERIFY approval status before proceeding. Once confirmed, INITIATE autonomous execution mode. STOP IMMEDIATELY upon detecting ANY requirement changes.
|
|
@@ -118,9 +119,13 @@ After all task cycles finish, run verification agents **in parallel** before the
|
|
|
118
119
|
2. **Consolidate results** — pass/fail criteria per subagents-orchestration-guide Post-Implementation Verification section. Present unified verification report to user.
|
|
119
120
|
|
|
120
121
|
3. **Fix cycle** (when any verifier failed, max 2 cycles):
|
|
121
|
-
-
|
|
122
|
-
-
|
|
123
|
-
|
|
122
|
+
- Create a consolidated fix task file (e.g., `docs/plans/tasks/post-impl-fixes-YYYYMMDD.md`) using the task-template; populate Target Files with the union of file paths referenced by all verifiers' `requiredFixes[].location` / `discrepancies[].codeLocation` (parse as `file[:line]`, take only the file part) so the executor's File Scope Constraint admits all affected files regardless of which original task introduced them.
|
|
123
|
+
- **Normalize verifier outputs** into a unified `requiredFixes[]` before invoking task-executor-frontend:
|
|
124
|
+
- `security-reviewer.requiredFixes[]` (already `{location, issue, fix}`) → pass through as-is.
|
|
125
|
+
- `code-verifier.discrepancies[]` → convert each actionable discrepancy (status `drift` / `gap` / `conflict`) to `{location: discrepancy.codeLocation, issue: discrepancy.claim, fix: "[specific correction needed to restore Design Doc consistency, derived from discrepancy.classification and evidence]"}`.
|
|
126
|
+
- When a `discrepancy.codeLocation` is `null` (claim is unimplemented), set `location` to the planned target file path and add that file to the consolidated task's Target Files. If no target file can be determined, escalate to user instead of invoking Fix Mode.
|
|
127
|
+
- Invoke task-executor-frontend in **Fix Mode** with `task_file` set to the consolidated path and `requiredFixes` set to the normalized array.
|
|
128
|
+
- Then quality-fixer-frontend, then re-run only the failed verifiers.
|
|
124
129
|
- If still failing after 2 cycles → Escalate to user with remaining findings
|
|
125
130
|
|
|
126
131
|
4. **All passed** → Proceed to completion report
|
|
@@ -106,7 +106,7 @@ Invoke task-executor-frontend using Agent tool:
|
|
|
106
106
|
Invoke quality-fixer-frontend using Agent tool:
|
|
107
107
|
- `subagent_type`: "quality-fixer-frontend"
|
|
108
108
|
- `description`: "Quality gate check"
|
|
109
|
-
- `prompt`: "Confirm quality gate passage for fixed files."
|
|
109
|
+
- `prompt`: "Confirm quality gate passage for fixed files. task_file: docs/plans/tasks/review-fixes-YYYYMMDD.md. filesModified: [extract from the prior implementation step's response]."
|
|
110
110
|
|
|
111
111
|
### Step 9: Re-validate code-reviewer
|
|
112
112
|
|
|
@@ -151,3 +151,14 @@ Remaining issues:
|
|
|
151
151
|
- Committed secrets (blocked → human intervention)
|
|
152
152
|
|
|
153
153
|
**Scope**: Design Doc compliance validation, security review, and auto-fixes.
|
|
154
|
+
|
|
155
|
+
## Scope Boundary for Subagents
|
|
156
|
+
|
|
157
|
+
Append the following block to every subagent prompt invoked from this recipe:
|
|
158
|
+
|
|
159
|
+
```
|
|
160
|
+
Scope boundary for subagents:
|
|
161
|
+
Operate within the task scope and referenced files in the prompt.
|
|
162
|
+
Use loaded skills to execute that scope.
|
|
163
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
164
|
+
```
|
|
@@ -39,7 +39,7 @@ When user responds to questions:
|
|
|
39
39
|
|
|
40
40
|
### 5. After Scale Determination: Register All Flow Steps with TaskCreate (Required)
|
|
41
41
|
|
|
42
|
-
After scale determination, **register all steps of the applicable subagents-orchestration-guide skill flow with TaskCreate**. Always include
|
|
42
|
+
After scale determination, **register all steps of the applicable subagents-orchestration-guide skill flow with TaskCreate**. Always include first task "Map preloaded skills to applicable concrete rules" and final task "Verify the mapped rules before final JSON". After registration, proceed through the flow referencing TaskList.
|
|
43
43
|
|
|
44
44
|
### 6. Execute Next Action
|
|
45
45
|
|
|
@@ -58,9 +58,18 @@ After scale determination, **register all steps of the applicable subagents-orch
|
|
|
58
58
|
|
|
59
59
|
**Flow Adherence**: Follow "Autonomous Execution Task Management" in subagents-orchestration-guide skill, managing 4 steps with TaskCreate/TaskUpdate
|
|
60
60
|
|
|
61
|
-
##
|
|
61
|
+
## Scope Boundary for Subagents
|
|
62
62
|
|
|
63
|
-
|
|
63
|
+
Append the following block to every subagent prompt invoked from this recipe:
|
|
64
|
+
|
|
65
|
+
```
|
|
66
|
+
Scope boundary for subagents:
|
|
67
|
+
Operate within the task scope and referenced files in the prompt.
|
|
68
|
+
Use loaded skills to execute that scope.
|
|
69
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
70
|
+
```
|
|
71
|
+
|
|
72
|
+
Additionally, include the following constraint at the end of every sub-agent prompt, as rule-advisor invocation from sub-agents causes system crash:
|
|
64
73
|
```
|
|
65
74
|
[Constraint] rule-advisor can only be used by Main AI
|
|
66
75
|
```
|
|
@@ -73,11 +82,11 @@ Following "Autonomous Execution Task Management" in subagents-orchestration-guid
|
|
|
73
82
|
2. **CHECK task-executor response**:
|
|
74
83
|
- `status: "escalation_needed"` or `"blocked"` → STOP and escalate to user
|
|
75
84
|
- `requiresTestReview` is `true` → Execute **integration-test-reviewer**
|
|
76
|
-
- `needs_revision` → Return to step 1
|
|
85
|
+
- `needs_revision` → Return to step 1 and re-invoke task-executor in **Fix Mode** by passing the same `task_file` and the `requiredFixes[]` array as input
|
|
77
86
|
- `approved` → Proceed to step 3
|
|
78
87
|
- Otherwise → Proceed to step 3
|
|
79
|
-
3. **INVOKE quality-fixer**: Execute all quality checks and fixes (cross-layer: see Layer-Aware Agent Routing). **Always pass** the current task file path as `task_file`
|
|
80
|
-
- `stub_detected` → Return to step 1
|
|
88
|
+
3. **INVOKE quality-fixer**: Execute all quality checks and fixes (cross-layer: see Layer-Aware Agent Routing). **Always pass** the current task file path as `task_file` and the implementation step's `filesModified` array as `filesModified` (this scopes the stub-detection step to the task's actual write set; without it, quality-fixer falls back to `git diff HEAD`)
|
|
89
|
+
- `stub_detected` → Return to step 1 and re-invoke task-executor in **Fix Mode** by passing the same `task_file` and the `incompleteImplementations[]` array as input
|
|
81
90
|
- `blocked` → Escalate to user
|
|
82
91
|
- `approved` → Proceed to step 4
|
|
83
92
|
4. **COMMIT on approval**: Execute git commit
|
|
@@ -88,7 +97,7 @@ After all task cycles finish, invoke security-reviewer before the completion rep
|
|
|
88
97
|
1. **Agent tool** (subagent_type: "security-reviewer") → Pass Design Doc path and implementation file list
|
|
89
98
|
2. Check response:
|
|
90
99
|
- `approved` or `approved_with_notes` → Proceed to completion report (include notes if present)
|
|
91
|
-
- `needs_revision` →
|
|
100
|
+
- `needs_revision` → Create a consolidated fix task file (`docs/plans/tasks/security-fixes-YYYYMMDD.md`) using the task-template; populate Target Files with the union of file paths referenced by `requiredFixes[].location` (parsed as `file[:line]`, take only the file part) so the executor's File Scope Constraint admits all affected files regardless of which original task introduced them. Then invoke task-executor in **Fix Mode** with `task_file` set to the new consolidated path and `requiredFixes` set to the security-reviewer array, followed by quality-fixer, then re-invoke security-reviewer.
|
|
92
101
|
- `blocked` → Escalate to user
|
|
93
102
|
|
|
94
103
|
### Test Information Communication
|
|
@@ -106,7 +106,7 @@ Invoke task-executor using Agent tool:
|
|
|
106
106
|
Invoke quality-fixer using Agent tool:
|
|
107
107
|
- `subagent_type`: "quality-fixer"
|
|
108
108
|
- `description`: "Quality gate check"
|
|
109
|
-
- `prompt`: "Confirm quality gate passage for fixed files."
|
|
109
|
+
- `prompt`: "Confirm quality gate passage for fixed files. task_file: docs/plans/tasks/review-fixes-YYYYMMDD.md. filesModified: [extract from the prior implementation step's response]."
|
|
110
110
|
|
|
111
111
|
### 9. Re-validate code-reviewer
|
|
112
112
|
|
|
@@ -152,3 +152,14 @@ Remaining issues:
|
|
|
152
152
|
- Committed secrets (blocked → human intervention)
|
|
153
153
|
|
|
154
154
|
**Scope**: Design Doc compliance validation, security review, and auto-fixes.
|
|
155
|
+
|
|
156
|
+
## Scope Boundary for Subagents
|
|
157
|
+
|
|
158
|
+
Append the following block to every subagent prompt invoked from this recipe:
|
|
159
|
+
|
|
160
|
+
```
|
|
161
|
+
Scope boundary for subagents:
|
|
162
|
+
Operate within the task scope and referenced files in the prompt.
|
|
163
|
+
Use loaded skills to execute that scope.
|
|
164
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
165
|
+
```
|
|
@@ -67,7 +67,12 @@ ls docs/ui-spec/*.md 2>/dev/null
|
|
|
67
67
|
|
|
68
68
|
### ステップ3: タスクファイル作成 [GATE]
|
|
69
69
|
|
|
70
|
-
|
|
70
|
+
**事前確認**: ステップ2 の各呼び出し結果について `generatedFiles.integration` を検査する:
|
|
71
|
+
- `integration` がパス → 該当レイヤーのタスク作成へ進む
|
|
72
|
+
- `integration` が `null` → 該当レイヤーのタスク作成をスキップし、レイヤー名と生成エージェントの `e2eAbsenceReason`(該当時)を最終レポート用に記録
|
|
73
|
+
- 全レイヤーが `integration: null` を返した → ステップ4〜7 を完全にスキップし、「どのレイヤーでも統合テストスケルトンは生成されなかった」と各レイヤーの理由を添えて報告して終了
|
|
74
|
+
|
|
75
|
+
`integration` パスが非nullのレイヤーごとに1つのタスクファイルを作成。monorepo-flow.mdの命名規則に従い、エージェントルーティングを決定的にする:
|
|
71
76
|
- バックエンドのDesign Doc → `docs/plans/tasks/integration-tests-backend-task-YYYYMMDD.md`
|
|
72
77
|
- フロントエンドのDesign Doc → `docs/plans/tasks/integration-tests-frontend-task-YYYYMMDD.md`
|
|
73
78
|
- 単一レイヤー(バックエンド確定) → `docs/plans/tasks/integration-tests-backend-task-YYYYMMDD.md`
|
|
@@ -80,22 +85,28 @@ name: [機能名]の[レイヤー]統合テスト実装
|
|
|
80
85
|
type: test-implementation
|
|
81
86
|
---
|
|
82
87
|
|
|
83
|
-
##
|
|
88
|
+
## Implementation Content
|
|
84
89
|
|
|
85
90
|
スケルトンファイルに定義されたテストケースを実装する。
|
|
86
91
|
|
|
87
|
-
##
|
|
92
|
+
## Target Files
|
|
88
93
|
|
|
89
94
|
- スケルトン: [ステップ2のgeneratedFilesからレイヤー別パス]
|
|
90
|
-
- Design Doc: [ステップ1のレイヤー別Design Doc]
|
|
91
95
|
|
|
92
|
-
##
|
|
96
|
+
## Investigation Targets
|
|
97
|
+
|
|
98
|
+
- Design Doc: [ステップ1のレイヤー別Design Doc] — AC マッピングと契約定義の参照用
|
|
99
|
+
|
|
100
|
+
## Investigation Notes
|
|
101
|
+
(実装観察事項を実装開始前にここへ追記する。)
|
|
102
|
+
|
|
103
|
+
## Implementation Steps
|
|
93
104
|
|
|
94
105
|
- [ ] スケルトンの各テストケースを実装
|
|
95
106
|
- [ ] 全テストがパスすることを確認
|
|
96
107
|
- [ ] カバレッジが要件を満たすことを確認
|
|
97
108
|
|
|
98
|
-
##
|
|
109
|
+
## Completion Criteria
|
|
99
110
|
|
|
100
111
|
- 全スケルトンテストケースが実装済み
|
|
101
112
|
- 全テストがパス
|
|
@@ -129,13 +140,13 @@ Agentツールでintegration-test-reviewerを呼び出す:
|
|
|
129
140
|
|
|
130
141
|
ステップ5の結果を確認:
|
|
131
142
|
- `status: approved` → 完了としてマーク、ステップ7へ進む
|
|
132
|
-
- `status: needs_revision` →
|
|
143
|
+
- `status: needs_revision` → ルーティング先の task-executor を **Fix Mode** で再起動。同じ `task_file` と `requiredFixes[]` を渡す(プロンプト詳細は下記)。その後ステップ5に戻る
|
|
133
144
|
|
|
134
145
|
タスクファイル名パターンでルーティングしてtask-executorを呼び出す:
|
|
135
146
|
- `*-backend-task-*` → `subagent_type`: "task-executor"
|
|
136
147
|
- `*-frontend-task-*` → `subagent_type`: "task-executor-frontend"
|
|
137
148
|
- `description`: "レビュー指摘の修正"
|
|
138
|
-
- `prompt`: "
|
|
149
|
+
- `prompt`: "task_file: [ステップ4で使用したのと同じタスクファイルパス]。requiredFixes: [ステップ5の requiredFixes 配列]。Fix Mode を適用(タスクのチェックボックスはステップ4で既に `[x]` になっている)。"
|
|
139
150
|
|
|
140
151
|
### ステップ7: 品質チェック
|
|
141
152
|
|
|
@@ -143,16 +154,28 @@ Agentツールでintegration-test-reviewerを呼び出す:
|
|
|
143
154
|
- `*-backend-task-*` → `subagent_type`: "quality-fixer"
|
|
144
155
|
- `*-frontend-task-*` → `subagent_type`: "quality-fixer-frontend"
|
|
145
156
|
- `description`: "最終品質保証"
|
|
146
|
-
- `prompt`: "このワークフローで追加されたテストファイルの最終品質保証。全テストを実行しカバレッジを確認。"
|
|
157
|
+
- `prompt`: "このワークフローで追加されたテストファイルの最終品質保証。全テストを実行しカバレッジを確認。task_file: [タスクファイルパス]。filesModified: [直近の実装ステップのレスポンスから抽出 — 通常はステップ4、修正再実行が走った場合はステップ4とステップ6の和集合]。"
|
|
147
158
|
|
|
148
159
|
**期待される出力**: `status` (approved/stub_detected/blocked)
|
|
149
160
|
|
|
150
|
-
quality-fixer
|
|
151
|
-
- `stub_detected` → `incompleteImplementations[]
|
|
161
|
+
quality-fixer レスポンスをチェック:
|
|
162
|
+
- `stub_detected` → ステップ4 に戻り、同じ `task_file` と `incompleteImplementations[]` 配列を入力として task-executor を **Fix Mode** で再起動し、ステップ4→5→6→7 を再実行
|
|
152
163
|
- `blocked` → ユーザーにエスカレーション
|
|
153
|
-
- `approved` →
|
|
164
|
+
- `approved` → ステップ8 へ
|
|
154
165
|
|
|
155
166
|
### ステップ8: コミット
|
|
156
167
|
|
|
157
|
-
quality-fixer
|
|
158
|
-
- Bashで適切なメッセージを付けてテストファイルをコミット
|
|
168
|
+
quality-fixer から `approved` の場合:
|
|
169
|
+
- Bash で適切なメッセージを付けてテストファイルをコミット
|
|
170
|
+
|
|
171
|
+
## サブエージェントのスコープ境界
|
|
172
|
+
|
|
173
|
+
本レシピから呼び出すサブエージェントプロンプトの末尾に以下のブロックを必ず付与する:
|
|
174
|
+
|
|
175
|
+
```
|
|
176
|
+
Scope boundary for subagents:
|
|
177
|
+
Operate within the task scope and referenced files in the prompt.
|
|
178
|
+
New files derived from the requested deliverable are in scope (e.g., test skeleton files specified by the recipe).
|
|
179
|
+
Use loaded skills to execute that scope.
|
|
180
|
+
Escalate when the required fix or investigation falls outside that scope.
|
|
181
|
+
```
|