steward-orchestrator

Orchestrate LiteLLM A2A agent cycles with health monitoring and proposal tracking.

Updated Mar 24, 2026
One-click install
npx skills add https://github.com/Heretek-AI/heretek-skills --skill steward-orchestrator
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: steward-orchestrator
Source: https://github.com/Heretek-AI/heretek-skills/tree/main/skills/steward-orchestrator
Command: npx skills add https://github.com/Heretek-AI/heretek-skills --skill steward-orchestrator

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Oversees and steers the autonomous collective, ensuring health monitoring, workflow enforcement, deliberation coordination, proposal management, and deadlock resolution so the system remains productive and aligned with charter principles.

Core Features & Use Cases

  • Monitor health: check status of all agents via LiteLLM A2A and surface actionable health signals.
  • Enforce workflow: orchestrate intel delivery, deliberation, voting, implementation, and stewardship handoffs to prevent stalls.
  • Manage cycles and proposals: trigger intelligence cycles, question reviews, and escalate deadlocks; track ledger state and pending actions.
  • Use Case: In a multi-agent run, Steward automatically detects stalled deliberations, triggers Explorer intelligence, and coordinates Triad votes to push decisions forward.

Quick Start

Trigger a complete agent health check and orchestrate the next cycle across all Triad teams.

Frequently Asked Questions about steward-orchestrator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I orchestrate a multi-agent workflow and prevent stalled deliberations?

You can orchestrate multi-agent workflows by enforcing intel delivery, deliberation, voting, and implementation handoffs. This prevents stalls by automatically detecting stalled deliberations, triggering agent intelligence, and coordinating votes to push decisions forward.

How does agent health monitoring work with LiteLLM A2A endpoints?

Agent health monitoring checks the status of all agents via LiteLLM A2A endpoints and surfaces actionable health signals. This ensures the collective remains productive and aligned with charter principles during autonomous operations.

What is the best way to resolve deadlocks in an autonomous agent collective?

The best way to resolve deadlocks is by escalating them through triggered intelligence cycles and question reviews. The orchestrator tracks ledger state and pending actions to manage proposals and push stalled decisions forward.

Can I track proposal state and pending actions across multiple agents using a ledger?

Yes, you can track ledger-based proposals and pending actions across agents. The system triggers intelligence cycles, manages question reviews, and escalates deadlocks while maintaining the ledger state for governance.

Do I need LiteLLM A2A to coordinate agent cycles and stewardship handoffs?

Yes, LiteLLM A2A integration is required to coordinate agent cycles, monitor health, and enforce stewardship handoffs. The system relies on these endpoints to surface health signals and orchestrate the collective workflow.

Why does my multi-agent workflow stall during the deliberation phase?

Workflows stall during deliberation when votes are not coordinated or intelligence is not triggered. The orchestrator prevents this by enforcing the workflow, triggering Explorer intelligence, and coordinating Triad votes to push decisions forward.