agent-core-impl:omp.core@current

{
  "id": "agent-core-impl:omp.core@current",
  "_kind": "AgentCoreImpl",
  "_file": "agent-stack/core-impls/omp-core-current.yaml",
  "_cluster": "agent-stack",
  "attributes": {
    "displayName": "OMP Core (current)",
    "agentVersionId": "agent-version:omp@current",
    "packageRef": "source-ref:omp-github",
    "loopIteratorPolicy": "tool-use-loop",
    "loopIteratorNotes": "OMP inherits Pi's serial tool-use loop. Loop advances on provider end-of-turn;\nthinking content is always preserved (passthrough) for downstream display.\n",
    "contextManagementStrategy": "platform-managed",
    "compactionTriggerNotes": "Provider-managed; OMP exposes no compaction threshold.\n",
    "subagentInvokerPolicy": "tool-call-handoff",
    "subagentInvokerNotes": "Same as Pi: no first-class subagent primitive; sub-tasks are tool-call handlers\nthat may spawn a nested OMP session.\n",
    "resultSynthesisPolicy": "model-direct",
    "resultSynthesisNotes": "Final assistant text is streamed; thinking blocks are interleaved verbatim.\nSession ends on provider end-of-turn.\n",
    "stopDetectionStrategy": "structured-end-event",
    "transportClientLibrary": "omp-internal",
    "supportedTransportProtocols": [
      "model-transport:anthropic-messages",
      "model-transport:openai-responses"
    ],
    "parallelToolCallHandling": "sequentialized",
    "streamingFidelity": "full",
    "thinkingChannelHandling": "passthrough"
  },
  "outgoingEdges": [
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "layer:4-agent-core",
      "kind": "realizes",
      "attributes": {}
    },
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "agent:omp",
      "kind": "composes",
      "attributes": {
        "role": "core"
      }
    },
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "model-transport:anthropic-messages",
      "kind": "speaks",
      "attributes": {}
    },
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "model-transport:openai-responses",
      "kind": "speaks",
      "attributes": {}
    },
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "capability:streaming",
      "kind": "supports",
      "attributes": {
        "versionRange": "*",
        "level": "full"
      }
    },
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "capability:supports-thinking",
      "kind": "supports",
      "attributes": {
        "versionRange": ">=1.0.0",
        "level": "full",
        "notes": "OMP always supports thinking per task brief.\n"
      }
    },
    {
      "from": "agent-core-impl:omp.core@current",
      "to": "capability:supports-tool-use",
      "kind": "supports",
      "attributes": {
        "versionRange": "*",
        "level": "full"
      }
    }
  ],
  "incomingEdges": [
    {
      "from": "agent-version:omp@current",
      "to": "agent-core-impl:omp.core@current",
      "kind": "composed_of",
      "attributes": {
        "role": "core"
      }
    }
  ]
}