Delegation

Orchestrates parallel work through background agents, custom agents, and coordinated agent teams.

Updated Jul 10, 2026
One-click install
npx skills add https://github.com/Kaleb-Rupe/aurora --skill delegation-kaleb-rupe
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Delegation
Source: https://github.com/Kaleb-Rupe/aurora/tree/main/claude/skills/Utilities/Delegation
Command: npx skills add https://github.com/Kaleb-Rupe/aurora --skill delegation-kaleb-rupe

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Complex work often involves multiple independent workstreams that run slowly when executed sequentially. This Skill routes work to the right parallelization mechanism—built-in agents, background agents, custom composed agents, or persistent agent teams—so independent tasks execute concurrently instead of one at a time. ## Core Features & Use Cases - Multiple Delegation Patterns: Launch built-in specialized agents (Engineer, Architect, Explore, Plan), worktree-isolated agents for file-safe parallelism, background agents for non-blocking work, or foreground agents for blocking dependencies. - Agent Teams & Swarms: Create persistent coordinated teams via TeamCreate with shared task lists, message-based coordination, and multi-turn collaboration across teammates. - Two-Tier Delegation: Choose lightweight delegation (haiku model, 3-turn limit) for one-shot extraction or classification, and full delegation for multi-step tasks requiring tool use. - Use Case: When refactoring a codebase across 10 files, dispatch parallel Task calls with worktree isolation so each agent edits safely in its own working tree while results are collected concurrently. ## Quick Start Ask the assistant to parallelize this work by spinning up background agents for each independent workstream.

Frequently Asked Questions about Delegation

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

FAQPage Schema
How do I run multiple agents in parallel?

Create multiple Task calls in a single message, giving each agent one unit of work. Use run_in_background for non-blocking execution or worktree isolation when agents edit the same files, then collect results when all complete.

What is the difference between custom agents and agent teams?

Custom agents are one-shot parallel workers launched via Task with composed personalities and no shared state. Agent teams use TeamCreate for persistent coordination with shared task lists, messaging, and multi-turn collaboration.

When should I use background agents instead of foreground agents?

Use background agents with run_in_background set to true when results are not needed immediately, such as research, long builds, or parallel investigations. Use foreground agents when the result is required before proceeding.

When should I not delegate work to agents?

Avoid delegating tasks that Grep, Glob, or Read can complete in under two seconds, single-file changes, or workstreams fewer than three. Also avoid full delegation for one-shot extraction; use lightweight delegation with a haiku model instead.

How do worktree-isolated agents prevent file conflicts?

Setting isolation to worktree gives each agent its own git working tree, so parallel agents never conflict on the same files. The worktree is auto-created on spawn and auto-cleaned when the agent finishes unless changes were made.