II.
OrchestrationPrimitive JSON
Structured · liveorch-primitive:agent-failover
Agent Failover json
Inspect the normalized record payload exactly as the atlas UI reads it.
{
"id": "orch-primitive:agent-failover",
"_kind": "OrchestrationPrimitive",
"_file": "agent-stack/interaction-primitives/orchestration-primitives.yaml",
"_cluster": "agent-stack",
"attributes": {
"displayName": "Agent Failover",
"description": "Redirect a failing or unavailable agent's workload to a standby agent\nor alternate harness. When the primary agent is unhealthy, rate-limited,\nor exhausted after retries, agent-failover transparently re-assigns the\ntask to the next eligible agent from the fallback roster. Preserves\nworkspace state so the standby agent resumes from the last known-good\ncheckpoint rather than restarting from scratch.\n",
"orchestrationKind": "dispatch",
"trigger": "on-failure"
},
"outgoingEdges": [
{
"from": "orch-primitive:agent-failover",
"to": "layer:13-orchestration",
"kind": "realizes",
"attributes": {}
},
{
"from": "orch-primitive:agent-failover",
"to": "topic:circuit-breakers",
"kind": "orch_applies_to",
"attributes": {}
},
{
"from": "orch-primitive:agent-failover",
"to": "topic:agent-dispatch",
"kind": "orch_applies_to",
"attributes": {}
},
{
"from": "orch-primitive:agent-failover",
"to": "skill-area:multi-agent-coordination",
"kind": "orch_requires_skill_area",
"attributes": {}
},
{
"from": "orch-primitive:agent-failover",
"to": "skill-area:operations-automation",
"kind": "orch_requires_skill_area",
"attributes": {}
},
{
"from": "orch-primitive:agent-failover",
"to": "orch-primitive:agent-dispatch",
"kind": "orch_triggers",
"attributes": {}
},
{
"from": "orch-primitive:agent-failover",
"to": "orch-primitive:workspace-snapshot",
"kind": "orch_triggers",
"attributes": {}
}
],
"incomingEdges": [
{
"from": "agent-core-impl:droid.core@current",
"to": "orch-primitive:agent-failover",
"kind": "supports_orchestration_primitive",
"attributes": {}
}
]
}