polyphony

Provision per-task Docker workspaces with dedicated git branches for parallel agent execution.

705|56|Updated Dec 26, 2025
One-click install
npx skills add https://github.com/alinaqi/maggy --skill polyphony
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: polyphony
Source: https://github.com/alinaqi/maggy/tree/main/skills/polyphony
Command: npx skills add https://github.com/alinaqi/maggy --skill polyphony

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Polyphony prevents agent-to-agent conflicts and flaky results by giving each routed task its own isolated Docker workspace with a dedicated git branch.

Core Features & Use Cases

  • Container-isolated parallel agents: Run multiple sessions concurrently without shared working directories, ensuring clean tests and PR-ready branch output.
  • Policy-based routing: Evaluate tasks with complexity scoring and route them to the best available agent(s) using top-down match rules.
  • Workspace lifecycle management: Provision per-task clones, mount credentials safely as read-only overlays, verify outcomes, and clean up completed workspaces.

Use case: You need to fix a high-risk security bug and also delegate documentation research—Polyphony can route each activity to different agents while keeping their repositories and test runs fully isolated.

Quick Start

Use the Polyphony skill to spawn a task by telling your AI: "Run polyphony spawn 'Fix auth middleware timing attacks' and show routing decisions and final landed outcome."

Frequently Asked Questions about polyphony

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

FAQPage Schema
How do I run multiple AI agents in isolated Docker workspaces?

You can run multiple AI agents concurrently by provisioning per-task Docker workspaces with dedicated git branches, preventing agent-to-agent conflicts and ensuring clean test runs.

What is the best way to route tasks across different agent CLIs?

Policy-based routing evaluates tasks using complexity scoring and top-down match rules to route activities to the best available agent CLIs like Claude, Codex, or Kimi for parallel execution.

How do I manage credentials safely when running parallel agents?

Credential identity resolution mounts credentials safely as read-only volume overlays and applies environment overlays, isolating access per workspace during the orchestrator loop.

Do I need Docker or OrbStack to run multi-agent git branch orchestration?

Yes, Docker or OrbStack availability is required to provision isolated sandboxes, mount credentials, and execute the orchestrator loop that discovers, claims, routes, verifies, and lands results.

How does workspace lifecycle management handle completed agent tasks?

Workspace lifecycle management provisions per-task clones, verifies outcomes through staged CI verification, and cleans up completed workspaces after landing results to prevent environment pollution.