agent-decision-guide

Guide Skill versus Agent selection in GitHub Copilot CLI workflows.

2|Updated Feb 25, 2026

Search Tags:#workflow design #automation strategy #ai architecture #llm agents #copilot cli #skill vs agent

Ready to run

npx skills add https://github.com/elihuvillaraus/skills --skill agent-decision-guide

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-decision-guide
Source: https://github.com/elihuvillaraus/skills/tree/main/agent-decision-guide
Command: npx skills add https://github.com/elihuvillaraus/skills --skill agent-decision-guide

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill provides a clear framework to decide whether to use a Skill (knowledge injection) or an Agent (autonomous worker) within GitHub Copilot CLI, preventing confusion and ensuring efficient AI workflow design.

Core Features & Use Cases

Decision Framework: Offers a quick TL;DR rule and a detailed decision tree.
Role Clarification: Differentiates between Skills (informing the main AI) and Agents (separate subprocesses).
Use Case: Before designing a new automation, consult this Skill to determine if your task requires injecting knowledge into the current AI context (Skill) or delegating an autonomous, parallelizable task to a separate process (Agent).

Quick Start

Consult the agent-decision-guide skill to understand when to use a skill versus an agent.

Frequently Asked Questions about agent-decision-guide

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

FAQPage Schema

What is the difference between a skill and an agent in LLM workflows?▼

The difference between a skill and an agent is that a skill injects knowledge into the main AI context, while an agent operates as a separate autonomous subprocess for parallelizable tasks within GitHub Copilot CLI.

How do I decide whether to use an agent or a skill for AI automation?▼

To decide whether to use an agent or a skill for AI automation, consult the decision framework: use skills to inform the current AI context, and delegate to agents when you need an autonomous, parallelizable subprocess.

When should I use an agent instead of a skill in GitHub Copilot CLI?▼

You should use an agent instead of a skill in GitHub Copilot CLI when your workflow design requires delegating an autonomous task to a separate process, rather than simply injecting knowledge into the existing AI context.

Can I inject external knowledge into an autonomous LLM agent?▼

Injecting external knowledge into an autonomous LLM agent is handled by using a skill, which informs the main AI context directly, whereas agents function as separate subprocesses rather than knowledge injection mechanisms.

Does this decision guide classify existing AI tools by their operational mode?▼

Yes, this decision guide classifies existing skills and agents based on their intended operational mode, offering a detailed decision tree to clarify architectural differences and use cases for your automations.