houmao-agent-loop-generic

Plan typed local-close and relay loop runs for Houmao agent teams.

18|3|Updated Mar 5, 2026
One-click install
npx skills add https://github.com/igamenovoer/houmao --skill houmao-agent-loop-generic
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: houmao-agent-loop-generic
Source: https://github.com/igamenovoer/houmao/tree/main/src/houmao/agents/assets/system_skills/legacy/houmao-agent-loop-generic
Command: npx skills add https://github.com/igamenovoer/houmao --skill houmao-agent-loop-generic

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It converts a user’s intent about coordinating multiple Houmao agents into a single, typed generic loop plan and provides run-control actions (start, status, stop) for operating that accepted plan safely.

Core Features & Use Cases

  • Typed loop planning from intent: Decomposes a multi-agent communication graph into explicit local-close (pairwise) components and relay components with normalized component dependencies and policy.
  • Run control for the accepted plan: Supports operating an existing generic loop run via start, status, and stop while keeping the user-controlled agent outside the execution loop.
  • Composed graph guidance routed to owning skills: Generates the final generic control graph and routes execution semantics to Houmao-owned skills that handle messaging, gateways, mailbox follow-up, inspection, and elemental local-close/relay execution patterns.
  • Constraints and guardrails: Encodes delegation/forwarding restrictions, result routing contracts, completion evaluation location, and default stop posture (interrupt-first) to prevent ambiguous or unsafe orchestration.

Quick Start

Use the houmao-agent-loop-generic skill to turn your multi-agent goal into a typed generic loop plan and then start or stop the resulting run by sending the normalized run control to the designated Houmao master/root owner.

Frequently Asked Questions about houmao-agent-loop-generic

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

FAQPage Schema
How do I plan and control multi-agent communication graphs for execution?

Loop planning for multi-agent orchestration decomposes a communication graph into local-close and relay components with explicit dependencies. It keeps the user-controlled agent outside execution while routing operational control to designated owner skills for messaging and gateways.

What is a local-close and relay loop in agent orchestration?

Local-close and relay loops are multi-agent execution patterns that decompose communication graphs into pairwise and relay components. They enforce normalized dependencies, graph policy, and result routing contracts to maintain safe agent coordination.

How do I start and stop a multi-agent loop run safely?

You start and stop multi-agent loop runs by sending normalized run-control actions to the designated run owner. The accepted typed plan supports start, status, and stop operations while maintaining an interrupt-first default stop posture to prevent unsafe orchestration.

Can I keep my user-controlled agent outside the multi-agent execution loop?

Yes, you can keep a user-controlled agent outside the execution loop. The run owner decomposes the communication graph into typed component plans, routing execution guidance to designated owner skills while the user agent retains operational control.

What guardrails exist for agent result routing and delegation?

Guardrails for agent result routing include delegation and forwarding restrictions, result routing contracts, and defined completion evaluation locations. These constraints encode an interrupt-first default stop posture to prevent ambiguous or unsafe multi-agent orchestration.