{
  "name": "DN Managed Runtime Evidence Dataset",
  "version": "0.1",
  "publishedAt": "2026-10-19",
  "status": "documentation-baseline-no-runtime-benchmark",
  "license": "CC-BY-4.0",
  "licenseScope": "DN original observations and test descriptions only; linked vendor documentation retains its own rights.",
  "methodologyUrl": "https://decentralised.news/dn-managed-runtime-dataset-v0-1",
  "records": [
    {
      "id": "openai-agents",
      "name": "OpenAI Agents API / SDK",
      "evidenceType": "documentation-observation",
      "reviewedAt": "2026-10-19",
      "source": "https://developers.openai.com/api/docs/guides/agents",
      "observation": "Documentation separates managed agent sessions from application-controlled SDK execution.",
      "runtimeExecution": "not-executed",
      "modelVersion": null,
      "harnessVersion": null,
      "environment": null,
      "metrics": {
        "acceptedOutcomes": null,
        "taskAttempts": null,
        "costPerAcceptedOutcome": null,
        "coldStartMs": null,
        "duplicateEffects": null
      },
      "limitations": [
        "No provider account deployed or benchmark executed.",
        "Documentation does not establish measured reliability, safety or migration performance."
      ]
    },
    {
      "id": "aws-bma",
      "name": "Amazon Bedrock Managed Agents powered by OpenAI",
      "evidenceType": "documentation-observation",
      "reviewedAt": "2026-10-19",
      "source": "https://docs.aws.amazon.com/bedrock/latest/userguide/bedrock-managed-agents-openai.html",
      "observation": "AWS labels BMA public preview and separates conversation and execution-environment lifecycle.",
      "runtimeExecution": "not-executed",
      "modelVersion": null,
      "harnessVersion": null,
      "environment": null,
      "metrics": {
        "acceptedOutcomes": null,
        "taskAttempts": null,
        "costPerAcceptedOutcome": null,
        "coldStartMs": null,
        "duplicateEffects": null
      },
      "limitations": [
        "No provider account deployed or benchmark executed.",
        "Documentation does not establish measured reliability, safety or migration performance."
      ]
    },
    {
      "id": "google-adk-runtime",
      "name": "Google ADK / managed Agent Runtime",
      "evidenceType": "documentation-observation",
      "reviewedAt": "2026-10-19",
      "source": "https://adk.dev/deploy/agent-runtime/",
      "observation": "ADK deployment documentation describes sending application code and dependencies to a managed runtime.",
      "runtimeExecution": "not-executed",
      "modelVersion": null,
      "harnessVersion": null,
      "environment": null,
      "metrics": {
        "acceptedOutcomes": null,
        "taskAttempts": null,
        "costPerAcceptedOutcome": null,
        "coldStartMs": null,
        "duplicateEffects": null
      },
      "limitations": [
        "No provider account deployed or benchmark executed.",
        "Documentation does not establish measured reliability, safety or migration performance."
      ]
    },
    {
      "id": "microsoft-foundry",
      "name": "Microsoft Foundry Agent Service",
      "evidenceType": "documentation-observation",
      "reviewedAt": "2026-10-19",
      "source": "https://learn.microsoft.com/en-us/azure/foundry/agents/overview",
      "observation": "Documentation distinguishes prompt-agent configuration from hosted custom-code agents.",
      "runtimeExecution": "not-executed",
      "modelVersion": null,
      "harnessVersion": null,
      "environment": null,
      "metrics": {
        "acceptedOutcomes": null,
        "taskAttempts": null,
        "costPerAcceptedOutcome": null,
        "coldStartMs": null,
        "duplicateEffects": null
      },
      "limitations": [
        "No provider account deployed or benchmark executed.",
        "Documentation does not establish measured reliability, safety or migration performance."
      ]
    },
    {
      "id": "langgraph-self-hosted",
      "name": "Self-hosted LangGraph reference",
      "evidenceType": "documentation-observation",
      "reviewedAt": "2026-10-19",
      "source": "https://docs.langchain.com/oss/python/langgraph/persistence",
      "observation": "Documentation distinguishes thread checkpoints from cross-thread stores.",
      "runtimeExecution": "not-executed",
      "modelVersion": null,
      "harnessVersion": null,
      "environment": null,
      "metrics": {
        "acceptedOutcomes": null,
        "taskAttempts": null,
        "costPerAcceptedOutcome": null,
        "coldStartMs": null,
        "duplicateEffects": null
      },
      "limitations": [
        "No provider account deployed or benchmark executed.",
        "Documentation does not establish measured reliability, safety or migration performance."
      ]
    }
  ],
  "testSpecification": [
    {
      "id": "RT-01",
      "name": "Bounded completion",
      "expectedEvidence": "An independently checked accepted outcome under pinned task and policy."
    },
    {
      "id": "RT-02",
      "name": "Lost response after commit",
      "expectedEvidence": "Stable action identifier, authoritative target state and absence or presence of duplicate effects."
    },
    {
      "id": "RT-03",
      "name": "Scope narrowing",
      "expectedEvidence": "A prohibited child request is rejected without an external effect."
    },
    {
      "id": "RT-04",
      "name": "Revocation",
      "expectedEvidence": "A timestamped grant is rejected after its declared effective revocation boundary."
    },
    {
      "id": "RT-05",
      "name": "State transfer",
      "expectedEvidence": "A resumed synthetic task preserves required state and authority with a documented mapping."
    },
    {
      "id": "RT-06",
      "name": "Full cost accounting",
      "expectedEvidence": "Spend for all attempts, human work and an independently graded accepted-outcome denominator."
    }
  ],
  "benchmarkRuns": [],
  "localValidation": {
    "scope": "Data contract and fixture validation only, not provider execution.",
    "receipt": "/data/managed-agent-runtime/validation-v0.1.json"
  },
  "notice": "No provider scores, winner or safety certification. Null means unmeasured, never zero."
}