# 研究タスクを提出（デフォルトの“standard”戦略）
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Latest breakthroughs in quantum error correction",
    "context": { "force_research": true }
  }'

# 使用可能なSSEストリームURLを取得して提出（推奨）
# ヒント: force_researchを含めてResearchWorkflowにルーティング（さもなければルーターがSupervisorを選択する可能性あり）
curl -s -X POST http://localhost:8080/api/v1/tasks/stream \
  -H "Content-Type: application/json" \
  -d '{
        "query": "What are the main transformer architecture trends in 2025?",
        "context": { "force_research": true }
      }' | jq

# /tasks/stream呼び出しの後、レスポンスから"stream_url"を使用:
curl -N "http://localhost:8080/api/v1/stream/sse?workflow_id=task-..."

# 例: 基本的な概念クエリがReactをトリガー
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What is quantum computing?",
    "context": { "force_research": true }
  }'

# 例: 複数の側面比較がParallelをトリガー
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Compare LangChain and AutoGen frameworks: architecture, pricing, community, performance",
    "context": { "force_research": true }
  }'

# 例: 論理的依存関係のあるクエリがHybridをトリガー
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Analyze company X market strategy, then evaluate competitive response, then forecast next moves",
    "context": { "force_research": true }
  }'

# 例: エリアごとの反復研究のためにreact_per_taskを有効化
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "Comprehensive analysis of transformer architecture evolution 2017-2025",
    "context": {
      "force_research": true,
      "react_per_task": true,
      "react_max_iterations": 6
    },
    "research_strategy": "deep"
  }'

# クイックストラテジー（最小限の反復）
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "量子誤り訂正とは何ですか？",
    "context": { "force_research": true },
    "research_strategy": "quick"
  }'

# オーバーライド付きのディープストラテジー
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "LangChainとAutoGenフレームワークの比較",
    "context": {
      "force_research": true,
      "react_per_task": true,
      "react_max_iterations": 6
    },
    "research_strategy": "deep",
    "enable_verification": true
  }'

# アカデミックストラテジー（複雑さが0.7を超えると自動的にreact_per_taskを有効化）
curl -X POST http://localhost:8080/api/v1/tasks \
  -H "Content-Type: application/json" \
  -d '{
    "query": "現代NLPアーキテクチャにおける注意メカニズムの調査",
    "context": { "force_research": true },
    "research_strategy": "academic"
  }'

python -m shannon.cli --base-url http://localhost:8080 \
  submit "最新の量子コンピューティングのブレークスルー" \
  --force-research \
  --research-strategy deep --enable-verification

from shannon import ShannonClient

client = ShannonClient(base_url="http://localhost:8080")
handle = client.submit_task(
    "LangChainとAutoGenフレームワークの比較",
    context={
        "force_research": True,
        "research_strategy": "deep",
        "react_max_iterations": 6,
        "enable_verification": True,
    },
)
final = client.wait(handle.task_id)
print(final.result)
client.close()

{
  "task_id": "task-00000000-0000-0000-0000-000000000000",
  "status": "TASK_STATUS_COMPLETED",
  "result": "... 最終合成テキスト ...",
  "metadata": {
    "citations": [
      {
        "url": "https://example.com/article",
        "title": "ソースタイトル",
        "source": "example.com",
        "quality_score": 0.92,
        "credibility_score": 0.85
      }
    ],
    "verification": {
      "total_claims": 10,
      "overall_confidence": 0.78,
      "supported_claims": 7,
      "unsupported_claims": [
        "十分な証拠がない主張テキスト ..."
      ],
      "claim_details": [
        {
          "claim": "具体的な主張テキスト ...",
          "confidence": 0.81,
          "supporting_citations": [1, 3],
          "conflicting_citations": []
        }
      ]
    }
  },
  "created_at": "2025-11-07T12:00:00Z",
  "updated_at": "2025-11-07T12:02:00Z",
  "usage": {
    "input_tokens": 1234,
    "output_tokens": 5678,
    "total_tokens": 6912,
    "estimated_cost": 0.0234
  },
  "model_used": "gpt-5-2025-08-07",
  "provider": "openai"
}

{
  "verification": {
    "total_claims": 10,
    "overall_confidence": 0.78,
    "supported_claims": 7,
    "unsupported_claims": ["Claim lacking evidence..."],
    "claim_details": [
      {
        "claim": "Transformer architecture introduced in 2017",
        "confidence": 0.95,
        "supporting_citations": [1, 3],
        "conflicting_citations": []
      }
    ]
  }
}