Multi-Agent Orchestration Kit

Coordinate multiple specialized agents working together to solve complex tasks.

Step 0 / 14

Plannerplanner

idle

Researcherresearcher

idle

Writerwriter

idle

Criticcritic

idle

Agent Communication Log

Run the simulation to see agent messages

Topology: hierarchical

Planner coordinates all agents, delegates tasks top-down, and aggregates results.

Agent Roles

Planner— Decomposes goals, coordinates agents

Researcher— Gathers and analyzes information

Writer— Produces output content

Critic— Reviews quality, provides feedback

Integration Code

import { createOrchestrator, Agent } from 'agent-tools-kit/reasoning'

const orchestrator = createOrchestrator({
  topology: 'hierarchical',
  agents: [
    new Agent({ role: 'planner', model: 'gpt-4o', systemPrompt: '...' }),
    new Agent({ role: 'researcher', model: 'gpt-4o', tools: ['web-search'] }),
    new Agent({ role: 'writer', model: 'gpt-4o' }),
    new Agent({ role: 'critic', model: 'gpt-4o' }),
  ],
  communication: {
    protocol: 'message-passing',  // or 'shared-blackboard'
    maxRounds: 10,
  },
  onMessage: (from, to, content) => {
    console.log(`[${from.role} → ${to.role}]: ${content}`)
  },
})

const result = await orchestrator.execute(
  'Write a comprehensive report on AI safety'
)
// result.output, result.agentLogs, result.messageHistory