github-copilot-customization-architecture

Designs and audits GitHub Copilot customization systems across instructions, prompts, skills, agents, hooks, MCP servers, and plugins.

Updated Aug 12, 2026
One-click install
npx skills add https://github.com/schattenspiegel/skill-foundry-skills --skill github-copilot-customization-architecture-schattenspiegel
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: github-copilot-customization-architecture
Source: https://github.com/schattenspiegel/skill-foundry-skills/tree/main/skills/github-copilot-customization-architecture
Command: npx skills add https://github.com/schattenspiegel/skill-foundry-skills --skill github-copilot-customization-architecture-schattenspiegel

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Teams customizing GitHub Copilot in VS Code often create overlapping instructions, prompt files, skills, agents, hooks, MCP servers, and plugins that collide, duplicate rules, or grant excessive tool access. This Skill provides a selection model and audit workflow to choose the smallest correct artifact for each requirement. ## Core Features & Use Cases - Artifact Selection Model: Routes each requirement to the right artifact type (instructions, prompt file, Agent Skill, custom agent, hook, MCP server, or plugin) based on activation mode and trust boundary. - Collision Auditing: Builds a collision map to remove duplicate or contradictory rules and enforces one owner per rule. - Least-Privilege Tool Assignment: Assigns minimal tool access and explicit trust steps for executable hooks, servers, and plugins. - Use Case: A repository has coding rules, a PR-preparation task, a security gate, and issue tracking needs. Use this Skill to map each need to instructions, a prompt file, a PreToolUse hook, and an MCP server respectively, then verify every artifact parses and activates correctly. ## Quick Start Audit my repository's current Copilot customizations and design an artifact map that assigns each requirement to the correct instructions, prompt, skill, agent, hook, MCP, or plugin artifact.

Frequently Asked Questions about github-copilot-customization-architecture

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

FAQPage Schema
How do I choose between Copilot instructions, prompt files, and Agent Skills?▼

Use instructions for conventions that apply to nearly every request, prompt files for tasks a person deliberately starts via slash command, and Agent Skills for multi-step capabilities with resources that Copilot must discover. Match the artifact to the activation mode your requirement needs.

When should I use a hook instead of Copilot instructions?▼

Use a hook when behavior must happen deterministically even if the model forgets, such as a security gate on PreToolUse events. Instructions are prose the model may overlook; hooks execute on lifecycle events but are a preview feature requiring explicit trust.

Where do Copilot customization files go in a VS Code workspace?▼

Instructions live in .github/copilot-instructions.md or .github/instructions/, prompts in .github/prompts/, skills in .github/skills/, agents in .github/agents/, hooks in .github/hooks/, and MCP config in .vscode/mcp.json. User-scope variants live in the VS Code profile or ~/.copilot/ paths.

How do I resolve conflicting Copilot customization rules?▼

Build a collision map of all artifacts, assign one owner per rule, and remove duplicate or contradictory rules rather than relying on instruction ordering. Split requirements whose activation, authority, or verification differs into separate artifacts.

What are the security risks of Copilot hooks and MCP servers?▼

Hooks can execute arbitrary commands and MCP servers introduce external code, data, and secrets into the session. Never silently enable them; assign least-privilege tools, require explicit trust steps, and never start MCP servers merely to validate configuration.