orchestration

name	orchestration
description	Use this skill when the user wants to coordinate multiple agents, launch a council for decision-making, invoke remote agents, or create multi-step workflows. Triggers include "launch a council", "multi-agent", "invoke remote agent", "agent-to-agent", "A2A", "coordinate agents", "deliberation", "consensus", or any request involving multiple AI agents working together.
metadata	{"author":"CorvidLabs","version":"1.0"}

Orchestration — Multi-Agent Coordination

Launch councils, invoke remote agents, and coordinate complex multi-agent workflows.

MCP Tools

• corvid_launch_council — Multi-agent deliberation with governance tiers
• corvid_invoke_remote_agent — Send tasks to A2A-compatible agents
• corvid_discover_agent — Discover agents via A2A Agent Card
• corvid_manage_workflow — Graph-based workflow orchestration

Council governance tiers

• Advisory — Discussion only, chairman synthesizes
• Simple majority — >50% agreement
• Supermajority — >66% agreement
• Unanimous — All must agree

Examples

Launch a council

Use corvid_launch_council:
  topic: "Payment system architecture"
  participants: ["corvid-agent", "security-agent"]
  governance: "simple_majority"

Invoke a remote agent

1. Use corvid_discover_agent for https://agent.example.com
2. Use corvid_invoke_remote_agent to send it a task

Notes

• Max invocation depth: 3 (prevents circular calls)
• Remote agents discovered via /.well-known/agent-card.json
• Council decisions are recorded on-chain