{
  "format": "agent-article-v0",
  "article_id": "EA-134",
  "canonical_url": "/tools/ea-134-langgraph",
  "page_type": "reference",
  "claim_ids": [],
  "evidence_ids": [
    "EVD-EA-134-URL-399CF6369BD3",
    "EA-134-R11-RESEARCH",
    "EA-134-R11-WEB-READBACK"
  ],
  "unknown_ids": [
    "EA-134-UNKNOWN-01",
    "EA-134-UNKNOWN-02",
    "EA-134-UNKNOWN-CURRENTNESS"
  ],
  "operational_object_ids": [],
  "provider": [],
  "models": [],
  "plans": [],
  "surfaces": [],
  "regions": [],
  "current_status": "UNKNOWN",
  "observation_status": "SPEC_ONLY",
  "freshness": "UNKNOWN",
  "last_verified": null,
  "supersedes": [],
  "superseded_by": [],
  "provenance": [
    "COMM-043-EA0134",
    "SRC-EA-134-URL-399CF6369BD3",
    "WEBBODY-EA0134-R8-53487C36"
  ],
  "article_sources": [
    {
      "id": "EVD-EA-134-URL-399CF6369BD3",
      "evidence_class": "RESEARCH",
      "title": "source_class=official project documentation; exact_title=LangGraph overview; author/org=LangChain Inc.; canonical_url=",
      "href": "https://docs.langchain.com/oss/python/langgraph/overview",
      "supports": [],
      "limit": "元の監査済み資料に記載された参照先です。URLの復元は、資料の最新性や全主張の支持を新たに検証したことを意味しません。",
      "origin_class": "SOURCE_PACKET_REFERENCE",
      "source_or_experiment_ids": [
        "SRC-EA-134-URL-399CF6369BD3"
      ],
      "supporting_locators": [
        "EVIDENCE_TABLE",
        "- source_class=official project documentation; exact_title=LangGraph overview; author/org=LangChain Inc.; canonical_url=https://docs.langchain.com/oss/python/langgraph/overview"
      ],
      "scope": "EVIDENCE_TABLE",
      "provenance_status": "SOURCE_PACKET_REFERENCE_PROJECTED"
    },
    {
      "id": "EA-134-R11-RESEARCH",
      "evidence_class": "RESEARCH",
      "title": "EA-134 監査済みresearch / canonical revision binding",
      "supports": [
        "- ESTABLISHED: 公式DocsはLangGraphを低レベルorchestrationとして説明し、deterministic stepとLLM stepの混在、durable execution、streaming、persistence、human-in-the-loop、memory、trace/debugを構成として記載している。\n- NOT_ESTABLISHED: docsの“production-ready”や機能存在は、全deploymentでの耐久性・再実行安全性、LLMの正しさ、human approvalの適切さ、SLA、LangSmith等のプラン・地域可用性を証明しない。"
      ],
      "limit": "canonical監査・revision bindingの存在を示す。外部ソースcurrentnessが本文またはcontrol-planeで確定していない場合は、currentnessをUNKNOWNのまま保持する。",
      "origin_class": "AUDITED_CANONICAL_RESEARCH_PACKET",
      "source_or_experiment_ids": [
        "COMM-043-EA0134"
      ],
      "supporting_locators": [
        "COMM-043-EA0134",
        "EA-134-R5-LEGACY-MIGRATION-20261008-R1",
        "content/staging/EA-134.md"
      ],
      "scope": "監査済みEA本文・canonical revision・既存Source Graph relationの範囲。欠損currentnessは推測で閉じない。",
      "provenance_status": "AUDIT_PASS_CANONICAL_BINDING"
    },
    {
      "id": "EA-134-R11-WEB-READBACK",
      "evidence_class": "IMPL",
      "title": "EA-134 Web body identity / canonical readback",
      "supports": [
        "canonical staging bodyのSHA-256一致とR9 canonical ledger readbackをProduction projectionの入力として検証する。"
      ],
      "limit": "body identity/readbackは本文claim自体の真偽や外部ソースの最新性を追加保証しない。",
      "checked_at": null,
      "origin_class": "IMPLEMENTATION_READBACK",
      "source_or_experiment_ids": [
        "WEBBODY-EA0134-R8-53487C36"
      ],
      "supporting_locators": [
        "WEBBODY-EA0134-R8-53487C36",
        "content/staging/EA-134.md",
        "EA-134-R5-LEGACY-MIGRATION-20261008-R1"
      ],
      "scope": "R9/R10 canonical readback後のProduction生成時body identity。",
      "provenance_status": "CANONICAL_READBACK_BODY_HASH_MATCH"
    }
  ],
  "research_sections": [
    {
      "heading": "WHY_THIS_MATTERS_TO_UPDATE",
      "text": "durable execution・streaming・human-in-loop・persistenceを持つ低レベルorchestrationとして、LLM agentの制御フローと外部実行の境界を現行仕様で更新する。"
    },
    {
      "heading": "SCOPE_ROUTING",
      "text": "Enterprise AI Core / implementation / governance / audit / operability。LangSmith等の周辺サービスは別availability gateで扱う。"
    },
    {
      "heading": "CLAIM_OR_RESEARCH_QUESTION",
      "text": "現行公式DocsがLangGraphの実行・状態・人間介入・観測をどこまで仕様として記述しているか。durabilityやhuman-in-loopを安全保証と誤認しない境界は何か。"
    },
    {
      "heading": "EVIDENCE_TABLE",
      "text": "- source_class=official project documentation; exact_title=LangGraph overview; author/org=LangChain Inc.; canonical_url=https://docs.langchain.com/oss/python/langgraph/overview; publication/update/access_date=current docs accessed 2026-09-13 JST; version/release/tag/commit/API revision=current OSS Python docs（ページ上に固定releaseなし）; supporting_locator=Overview / Install / Build a basic graph / Persistence / Human-in-the-loop / Production（low-level orchestration、durable execution、streaming、human-in-loop、deterministicとLLM steps、persistence/memory、LangSmith tracing/evaluation、standalone利用の記述）; evidence_label=VERIFIED_SPEC+PRODUCT_VOLATILE; relation=support+limit"
    },
    {
      "heading": "FINDINGS",
      "text": "- ESTABLISHED: 公式DocsはLangGraphを低レベルorchestrationとして説明し、deterministic stepとLLM stepの混在、durable execution、streaming、persistence、human-in-the-loop、memory、trace/debugを構成として記載している。\n- NOT_ESTABLISHED: docsの“production-ready”や機能存在は、全deploymentでの耐久性・再実行安全性、LLMの正しさ、human approvalの適切さ、SLA、LangSmith等のプラン・地域可用性を証明しない。"
    },
    {
      "heading": "LIMITS / COUNTEREVIDENCE / FAILURE ENVELOPE",
      "text": "checkpointer/backend、idempotency、side-effect ordering、resume設計を誤ると再実行で二重actionが起こり得る。stream/traceの欠損、状態肥大、長時間workflow、tool失敗、LLM hallucinationは別途扱う。オーケストレーション機能をagent品質へ外挿しない。"
    },
    {
      "heading": "AVAILABILITY_GATES",
      "text": "package version、Python/JS実装差、checkpointer・database・deployment backend、LangSmithの利用条件、model/provider API、human approver identity/RBAC、外部tool権限。"
    },
    {
      "heading": "IMPLEMENTATION_STATE",
      "text": "OSS/Python official documentation + PRODUCT_VOLATILE。Docs exampleは実装入口であり、production deployment・外部side effect・認証tenantでの実測は未実施。"
    },
    {
      "heading": "BENCHMARK_OR_EVAL",
      "text": "Docsにtracing/evaluationの説明はあるが、特定業務の独立benchmark・durability SLO・二重実行率は本Packetで確立していない。UNKNOWN。"
    },
    {
      "heading": "SECURITY / PRIVACY / AUDIT / COST / OPERABILITY",
      "text": "Security=human-in-loopをaction境界に置き、side effectはidempotency keyとdeny-by-defaultで制御。Privacy=state/checkpoint/traceのPII保持と削除。Audit=graph version・node・state transition・approval・tool resultを相関。Cost=resume/retry/trace/storageを含む。Operability=checkpoint restore、timeout、dead-letter、rollback、backend障害を試験。"
    },
    {
      "heading": "SAFE_WORDING_FOR_FUTURE_BOOK",
      "text": "「LangGraphの公式Docsは、LLM stepと決定的stepを組み合わせる低レベルorchestrationとして、persistence・durable execution・streaming・human-in-the-loop等を記載している。これらは制御基盤の仕様候補であり、LLMの正しさ、side effectの安全性、全deploymentのSLAを自動保証しない。」"
    },
    {
      "heading": "UPDATE_WORK_HANDOFF",
      "text": "Core実装章へ、graph/state/checkpoint/tool side effectを別claimとして引き継ぐ。次回は固定package/commit、checkpointerの再実行仕様、LangSmithのversion・data boundary、公開issue/incident、業務benchmarkを補う。"
    },
    {
      "heading": "OPEN_UNKNOWN",
      "text": "current release/commit、各backendのdurability、exactly-once相当の保証有無、human approval race、LangGraph/LangSmith境界、JS/Python parity、tenant availability。"
    },
    {
      "heading": "STOP_CONDITION",
      "text": "“durable”をexactly-once・データ完全性・安全な外部actionへ読み替えない。Docsの例やtrace表示を業務品質の証拠にしない。\nUNIT_CLOSE_STATUS: INDIVIDUAL_BOUNDED_CLOSE / EVIDENCE_RECORDED / UNKNOWN_PRESERVED\n\n---"
    },
    {
      "heading": "AUDITED_BOUNDARY",
      "text": "LangGraph durable execution is not exactly-once external execution; external side effects require idempotency; HITL is not authorization; OSS distinct from hosted entitlement/production guarantees."
    },
    {
      "heading": "OPEN_UNKNOWN",
      "text": "All UNKNOWN/PRODUCT_VOLATILE/provider- or organization-specific limits from the authority/audit remain open unless explicitly closed by the audited handoff."
    }
  ],
  "human_url": "/tools/ea-134-langgraph",
  "markdown_url": "/agent/article.md?path=%2Ftools%2Fea-134-langgraph",
  "json_url": "/agent/article.json?path=%2Ftools%2Fea-134-langgraph"
}