Issue起点のワークフロー全体を制御するエージェント。issue-intakeによるトリアージ、 impact-analysisによる影響分析、ai-led-onboardingによる文脈構築、eld-sense-task-decompositionに よるタスク分解を統合し、Issue→PR完了までを一貫して管理する。 使用タイミング: (1) 「Issue #N を対応して」、(2) 「このIssueを解決して」、 (3) Issue起点で作業を開始する時、(4) 複数スキルを連携させてIssue対応したい時
Issue-driven development workflow orchestrator that manages the entire lifecycle from triage to PR. Automatically executes intake, impact analysis, context building, and task decomposition in the optimal sequence based on issue classification and severity.
/plugin marketplace add CAPHTECH/claude-marketplace/plugin install caphtech-plugin@caphtech-marketplaceIssue起点の開発ワークフロー全体を制御し、適切なスキルを適切なタイミングで起動する。
Phase 1: Intake(トリアージ)
└→ issue-intake
- classification: Critical/Major/Minor/Enhancement/NeedsInfo
- severity_score + confidence
- uncertainty_flags → 次フェーズ判断
Phase 2: Context(文脈構築)
└→ ai-led-onboarding
└→ impact-analysis(classification >= Major時)
- 影響範囲の特定
- リスク評価
Phase 3: Uncertainty Resolution(不確実性解消)
└→ resolving-uncertainty(uncertainty_flags存在時)
- 再現手順確認
- 期待動作確認
- 環境情報収集
Phase 4: Task Decomposition(タスク分解)
└→ eld-sense-task-decomposition
- 実行可能なタスクに分解
- 依存関係の特定
Phase 5: Implementation(実装)
└→ issue-workflow-orchestrator で進捗管理
- 各タスクの実行
- 観測・検証
Phase 6: Review(レビュー)
└→ PR作成・レビュー
| classification | severity | 適用テンプレート |
|---|---|---|
| NeedsInfo | - | uncertainty_resolution先行 |
| Critical | 9-10 | emergency(Phase 2-6並列可) |
| Major | 6-8 | standard(全Phase順次) |
| Minor | 3-5 | standard_light(Phase 3省略可) |
| Enhancement | 1-2 | lightweight(Phase 2,3省略) |
| From | To | 条件 |
|---|---|---|
| Intake | Context | classification != NeedsInfo |
| Intake | Uncertainty | classification == NeedsInfo |
| Context | Task Decomposition | 影響範囲が明確 |
| Context | Uncertainty | unknowns_and_assumptions.unknowns が多い |
| Uncertainty | Task Decomposition | 不確実性が解消 |
| Task Decomposition | Implementation | タスク計画承認済み |
| Implementation | Review | 全タスク完了 |
# Issue情報取得
gh issue view <number> --json title,body,labels,comments
# issue-intake実行
# → classification, severity_score, confidence, uncertainty_flags
issue-intakeの結果から適用テンプレートを決定:
recommended_workflow を参照uncertainty_flags の有無を確認選択されたテンプレートに従い、各Phaseを順次または並列で実行。 各Phase完了時にartifactを次Phaseに引き継ぐ。
issue-workflow-orchestratorの状態機械に従い:
workflow_state:
issue_ref: "repo#123"
current_phase: "implementation"
template: "standard"
phase_results:
intake:
status: completed
classification: Major
severity_score: 7
confidence: 0.8
context:
status: completed
risk_level: medium
affected_files: 5
task_decomposition:
status: completed
tasks: 3
implementation:
status: in_progress
completed_tasks: 1
remaining_tasks: 2
next_action: "タスク2を実行"
blockers: []
| 検知条件 | 閾値 | アクション |
|---|---|---|
| confidence低下 | < 0.3 | uncertainty_resolution再実行 |
| 影響範囲拡大 | 想定の2倍超 | approval_point挿入 |
| セキュリティ懸念検出 | - | emergency昇格検討 |
| タスク完了率停滞 | 3時間進捗なし | ブロッカー分析 |
Use this agent when analyzing conversation transcripts to find behaviors worth preventing with hooks. Examples: <example>Context: User is running /hookify command without arguments user: "/hookify" assistant: "I'll analyze the conversation to find behaviors you want to prevent" <commentary>The /hookify command without arguments triggers conversation analysis to find unwanted behaviors.</commentary></example><example>Context: User wants to create hooks from recent frustrations user: "Can you look back at this conversation and help me create hooks for the mistakes you made?" assistant: "I'll use the conversation-analyzer agent to identify the issues and suggest hooks." <commentary>User explicitly asks to analyze conversation for mistakes that should be prevented.</commentary></example>