Orchestrator Agent (CORE)
Global Autonomous Execution Coordinator · POST /v1/agent/orchestrator/run

Category: Core Orchestration · Execution Model: Deterministic, Stateful, Sequential

Purpose

The Orchestrator Agent is the top-level autonomous controller responsible for executing United AI’s full closed-loop, multi-agent operating cycle. It enforces execution order, manages system state, coordinates inter-agent dependencies, and guarantees that discovery, supply generation, scheduling, and reinforcement learning occur as a single coherent pipeline.

Expected Inputs

  • runType string — default: weekly-discovery
    Controls the orchestration mode. Determines which agents are executed and how results are interpreted.

Outputs

  • HTTP Response
    {
      "status": "success | partial | failed",
      "agentsTriggered": "string[]",
      "errors": "Array<{ agent: string, message: string }>",
      "historyId": "ObjectId"
    }
  • Side Effect
    Creates OrchestrationHistory record
  • Side Effect
    Writes AgentMemory (via LearningAgent)
  • Side Effect
    Triggers writes across ~20 downstream collections

Responsibilities

Lifecycle

Downstream Dependencies

Guarantees

Failure Model

The Orchestrator is resilient by design. If an agent fails, the error is logged and execution continues where safe. The final orchestration status reflects whether the run was fully successful, partially successful, or failed to initialize.

Internal State

{
  "persistence": [
    "OrchestrationHistory",
    "AgentMemory"
  ],
  "guarantees": [
    "Execution order is fixed and versioned",
    "Agents never execute out of dependency order",
    "A failed agent does not invalidate prior successful agents"
  ]
}

Sample cURL


curl -X POST https://api.whook.ai/v1/agent/orchestrator/run \
  -H "Content-Type: application/json" \
  -d '{"runType":"weekly-discovery"}'