agent-mux

Orchestrate cross-engine AI dispatches across Codex, Claude, and Gemini using a single JSON contract.

35|8|Updated Feb 13, 2026
One-click install
npx skills add https://github.com/buildoak/agent-mux --skill agent-mux
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-mux
Source: https://github.com/buildoak/agent-mux/tree/main/skill
Command: npx skills add https://github.com/buildoak/agent-mux --skill agent-mux

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrates cross-engine AI dispatches across Codex, Claude, and Gemini to enable launching, steering, and recovering multi-model work through a single JSON contract.

Core Features & Use Cases

  • Unified cross-engine dispatch: drive workers on Codex, Claude, and Gemini with one contract.
  • Prompt-driven workers: worker identity is defined inside Markdown prompts for deterministic behavior.
  • Recovery and steering: supports timeouts, mid-flight steer, and cross-engine coordination for complex workflows.

Quick Start

Run a profile dispatch (e.g., lifter) to launch a worker across Codex/Claude/Gemini and inspect the resulting artifacts.

Frequently Asked Questions about agent-mux

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

FAQPage Schema
How do I coordinate multi-engine AI dispatches across Codex, Claude, and Gemini?

Cross-engine AI dispatch across Codex, Claude, and Gemini is orchestrated through a single JSON contract that launches workers, steers mid-flight executions, and coordinates multi-model workflows across different engines.

What is a JSON contract workflow for multi-model AI agents?

A JSON contract workflow for multi-model agents defines worker identity and deterministic behavior inside Markdown prompts, enabling a unified dispatch that drives multiple AI engines simultaneously.

Can I recover timed-out AI dispatches or steer workers mid-flight?

Yes, you can recover timed-out dispatches and steer mid-flight workers. The orchestration supports cross-engine coordination to handle timeouts and adjust complex workflows during execution.

How do I launch a worker across multiple AI engines using one contract?

You launch a worker across multiple AI engines by running a profile dispatch that uses one JSON contract, then inspect the resulting artifacts generated by the cross-engine execution.

Do I need a root SKILL.md with frontmatter to use cross-engine dispatch?

Yes, cross-engine dispatch requires a root SKILL.md with mandatory frontmatter containing name and description, which drives discovery and allows injecting resources per the workflow requirements.

What's the best way to handle complex multi-model workflows across different AI engines?

The best way to handle complex multi-model workflows is using a unified JSON contract that applies prompt-driven worker identities, enabling deterministic behavior and cross-engine coordination for execution and recovery.