skills.rest
jkf87
jkf87COMMUNITY
Explore all skills by this author →
GitHub

harness

Route tasks across AI providers with deterministic model selection and multi-account pooling.

61|21|Updated Mar 25, 2026
Search Tags:#routing#Codex#orchestration#OpenRouter#multi-provider#harness#OpenClaw
One-click install
npx skills add https://github.com/jkf87/ohmyclaw --skill harness-jkf87
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: harness
Source: https://github.com/jkf87/ohmyclaw/tree/main
Command: npx skills add https://github.com/jkf87/ohmyclaw --skill harness-jkf87

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jq, python3, and includes scripts (resource) components.

What problem does it solve?

OpenClaw Harness coordinates multiple AI agents across providers to enable cross-model orchestration and multi-account routing.

Core Features & Use Cases

  • Deterministic model routing via routing.json and select-model.sh for consistent task assignment.
  • Multi-account pool management with round-robin, cooldown, and fan-out to maximize throughput.
  • OMX-style composable verbs and bridge notifications to orchestrate Plan→Work→Review cycles with real-time alerts.

Quick Start

Install the harness and run '/ohmyclaw' to launch the HUD dashboard.

Frequently Asked Questions about harness

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

FAQPage Schema
How does multi-provider AI orchestration work for routing tasks across OpenAI Codex and OpenRouter?▼

Multi-provider AI orchestration uses a deterministic routing configuration to assign tasks to specific models like Z.ai GLM, OpenAI Codex, and OpenRouter, managing a multi-account pool with round-robin and cooldown to optimize throughput and reliability.

What is a Plan-Work-Review lifecycle for orchestrating AI agents across different providers?▼

A Plan-Work-Review lifecycle is an orchestration cycle that sequentially structures AI agent tasks into planning, working, and reviewing phases, coordinated via OMX-style composable verbs and real-time channel bridge notifications.

Do I need jq and Python to route tasks across multiple AI providers?▼

Yes, routing tasks across multiple AI providers requires jq for deterministic model selection and Python for runtime state handling to manage the multi-account pool and execute the orchestration lifecycle.

How do I set up multi-account pool management for AI agents to maximize throughput?▼

Multi-account pool management for AI agents is set up by configuring round-robin selection, cooldown periods, and fan-out distribution, which maximizes throughput and ensures reliability across multiple external model providers.

Can I coordinate OpenClaw agents across Z.ai GLM and external OpenRouter models?▼

Yes, you can coordinate OpenClaw agents across Z.ai GLM, OpenAI Codex, and external OpenRouter models by using a deterministic routing source to select models and orchestrate tasks across a managed multi-account pool.

Related Skills

Explore More Software Engineering Skills→
markus41markus41
1

llmintegration

Integrate Claude, GPT, Gemini, and Ollama APIs with a pluggable provider abstraction.

Community
Intermediate
mofa-orgmofa-org
11

observatory-report

Bridge gateway telemetry JSON into visual or text health reports via predefined style templates.

Official
Advanced
LinnanliLinnanli
1

team

Coordinate multiple AI agents on a shared task list via tmux sessions.

Community
Advanced
distribucionesfayosmoreno-sysdistribucionesfayosmoreno-sys

workflow-orchestration-patterns

Orchestrate distributed system workflows using Temporal saga, entity, and fan-out/fan-in patterns.

Community
Advanced