community-collections

Welcome to your prompt collection! This repository contains prompts, instructions, agents, skills, and collections to enhance your GitHub Copilot experience.

📖 Repository Structure

├── prompts/              # Task-specific prompts (.prompt.md)
├── instructions/         # Coding standards and best practices (.instructions.md)
├── agents/               # AI personas and specialized modes (.agent.md)
├── skills/               # Agent skills with bundled assets (SKILL.md)
├── collections/          # Curated collections (.collection.yml)
├── mcp-server/           # Optional: MCP server configuration
├── scripts/              # Validation and creation utilities
├── .vscode/              # VS Code settings and extensions
└── package.json          # Node.js dependencies

🔗 About AI Primitives Hub

This repository is part of the AI Primitives Hub ecosystem - a VS Code extension and framework for managing AI prompt collections for GitHub Copilot.

What is AI Primitives Hub?

AI Primitives Hub is a comprehensive platform that enables:

Discovery: Browse and install prompt collections from various sources
Management: Organize prompts, instructions, agents, and skills in one place
Collaboration: Share collections within your team or organization
Version Control: Track changes and maintain consistency across projects

Key Features

🚀 Marketplace UI: Visual browsing and installation of collections
📦 Bundle Management: Install, update, and uninstall prompt bundles
👥 Team Collaboration: Shared profiles and favorites
🔧 Development Tools: Scaffolding, validation, and publishing workflows
🌐 Multi-Source Support: GitHub repositories, local files, and custom sources

Getting Started with AI Primitives Hub

Install the Extension: Search "AI Primitives Hub" in the VS Code Marketplace
Browse Collections: Use the Hub Explorer to discover available content
Install & Use: One-click installation makes prompts available in Copilot Chat

📖 Learn More: AI Primitives Hub Documentation

🚀 Quick Start

1. Install Dependencies

This project uses @prompt-registry/collection-scripts from npmjs.com.

# Install dependencies
npm install

2. Validate Your Collections

npm run validate

Checks:

✅ Required fields (id, name, description)
✅ ID format (lowercase, hyphens only)
✅ File references exist
✅ Valid YAML syntax

4. Validate Your Skills

npm run skill:validate

Checks:

✅ SKILL.md frontmatter (name, description)
✅ Name matches folder name
✅ No duplicate skill names
✅ Bundled asset size limits (max 5MB per file)

5. Use with VS Code

The scaffold includes VS Code configuration:

Recommended Extensions (auto-prompted on first open):

redhat.vscode-yaml - YAML language support with schema validation

Auto-configured Features:

YAML schema validation for .collection.yml files
IntelliSense for collection properties
Real-time validation errors

6. Ensure that the GitHub runner label is correctly configured

open workflows/publish.yml
look for runs-on: ubuntu-latest
ensure you are using the runner label as per recommendations of your organisation
update the githubRunner value in your project's configuration if needed

7. (Optional) Enable MCP Servers

What is MCP? Model Context Protocol allows your collection to provide custom tools and context to GitHub Copilot.

Quick Setup:

Edit your collections/*.collection.yml file
Uncomment the mcp section
Choose from pre-built servers (time, filesystem, memory) or create your own

See mcp-server/README.md for detailed instructions.

8. Publish to GitHub

# Initialize git (if needed)
git init
git add .
git commit -m "Initial commit"

# Create repo on GitHub, then:
git remote add origin https://github.com/YOUR_USERNAME/YOUR_REPO.git
git branch -M main
git push -u origin main

9. Use with AI Primitives Hub Extension

Option A: Add as Source

Open VS Code Command Palette (Ctrl+Shift+P)
Run: "AI Primitives Hub: Add Source"
Select "Github Releases" or "Collection from GitHub repository"
Enter your repo URL: https://github.com/YOUR_USERNAME/YOUR_REPO

Option B: Pre-configured Default The AI Primitives Hub extension automatically includes the official github/awesome-copilot source. Once published, your collection will be available similarly.

🌐 Publishing to Hub and Profiles

What are Hubs and Profiles?

Hubs are centralized configurations that group related prompt sources. They provide:

Curated Content: Pre-selected collections for specific teams or domains
Easy Access: One-click installation of entire content sets
Organization: Logical grouping by department, technology, or purpose

Profiles are subsets of hub content that users can activate:

Shared Profiles: Curated by hub maintainers for common use cases
Favorites: Personal selections for daily work
Context Switching: Quick activation based on current task

Publishing Your Collection to a Hub

Step 1: Prepare Your Collection

Ensure your collection is ready for sharing:

# Validate everything
npm run validate
npm run skill:validate

# Test installation locally
# (Use AI Primitives Hub extension to add your repo as a source)

Step 2: Contact Hub Maintainers

Identify the Right Hub: Find a hub that matches your content domain
Review Hub Guidelines: Check contribution requirements and standards
Submit Your Collection: Share your repository URL and collection details

Step 3: Hub Integration Process

Hub maintainers will:

Review Content: Check quality, relevance, and compliance
Test Integration: Verify installation and functionality
Add to Hub: Include your collection in the hub configuration
Create Profile: Optionally add to curated profiles

Step 4: User Access

Once published, users can:

Install the Hub: Add the hub to their AI Primitives Hub
Browse Collections: Find your content in the hub catalog
Install Collections: One-click install to their Copilot environment
Use Profiles: Activate curated profiles containing your content

Creating Your Own Hub

For teams wanting to create private hubs:

Hub Configuration: Create a hub-config.yml file
Source Management: Add approved repositories and collections
Profile Curation: Create profiles for different use cases
Distribution: Share hub URL with team members

📖 Learn More: Hub Configuration Guide

🎯 Choosing the Right Agentic Primitive

When creating content for your collection, it's important to choose the right type of agentic primitive for your use case. Each primitive serves a specific purpose and has different characteristics.

Quick Decision Guide

🎯 Single Task? → Use a Prompt for specific, reusable instructions
📚 Team Standards? → Use Instructions for sharing best practices
🤖 Specialized Expertise? → Use an Agent for consistent AI personas
🛠️ Complex Workflows? → Use a Skill for multi-file functionality
📦 Related Content? → Use a Collection to organize primitives

Detailed Comparison

Primitive	Best For	Complexity	Example
Prompt	Quick, specific tasks	Low	"Generate unit tests"
Instruction	Team standards & guidelines	Medium	"React coding standards"
Agent	Specialized AI roles	Medium-High	"Security auditor persona"
Skill	Complex capabilities with assets	High	"API documentation generator"
Collection	Organizing related content	Variable	"React development kit"

Interactive Decision Tree

flowchart TD
    A[What do you want to create?] --> B{Specific task or general knowledge?}
    B -->|Specific task| C{Reusable behavior needed?}
    B -->|General knowledge| D{Team standards or best practices?}
    
    C -->|Yes| E{Multiple files/assets required?}
    C -->|No| F[Create Prompt]
    
    E -->|Yes| G[Create Skill]
    E -->|No| H[Create Agent]
    
    D -->|Yes| I[Create Instruction]
    D -->|No| J[Create Collection]

📖 Complete Guide: Agentic Primitives Documentation

📝 Creating Content

Prompts (`prompts/*.prompt.md`)

Task-specific instructions for Copilot.

# Generate Unit Tests

Create comprehensive unit tests for the current file.

## Instructions

1. Analyze the code structure
2. Generate test cases for all public methods
3. Include edge cases and error scenarios
4. Use the project's testing framework

## Tags

#testing #quality #automation

Instructions (`instructions/*.instructions.md`)

Coding standards that apply automatically.

# TypeScript Best Practices

## Guidelines

- Use explicit types, avoid `any`
- Prefer `const` over `let`
- Document public APIs with JSDoc

## Applies To

- `**/*.ts`
- `**/*.tsx`

Agents (`agents/*.agent.md`)

Agents are conversational AI assistants that define specialized personas for GitHub Copilot Chat. Each agent has specific expertise and guidelines for how it should respond.

# Code Reviewer

You are a senior code reviewer focused on quality and best practices.

## Expertise

- Code quality and maintainability
- Security vulnerabilities
- Performance optimization

## Guidelines

- Provide constructive feedback
- Suggest specific improvements
- Explain the reasoning

Agent Skills (`skills/<name>/SKILL.md`)

Domain-specific capabilities for Copilot following the Agent Skills specification.

Creating a New Skill:

npm run skill:create

This runs an interactive wizard that prompts for:

Name: lowercase letters, numbers, and hyphens (e.g., code-review)
Description: 10-1024 characters explaining what the skill does

Skill Structure:

skills/
└── my-skill/
    ├── SKILL.md        # Required: skill definition with frontmatter
    ├── example.py      # Optional: bundled asset
    └── schema.json     # Optional: bundled asset

SKILL.md Format:

---
name: my-skill
description: "A concise description of what this skill enables (10-1024 chars)"
---

# My Skill

Detailed instructions for how Copilot should use this skill.

## Capabilities

What this skill enables Copilot to do.

## Usage

When and how Copilot should apply this skill.

Bundled Assets:

Include supporting files (templates, schemas, examples) alongside SKILL.md
Maximum 5MB per file
Referenced in the markdown body

Validation:

npm run skill:validate

Collections (`collections/*.collection.yml`)

Group related items together.

id: typescript-development
name: TypeScript Development
description: Essential prompts and instructions for TypeScript projects
tags:
  - typescript
  - development
items:
  - path: prompts/generate-tests.prompt.md
    kind: prompt
  - path: instructions/typescript-style.instructions.md
    kind: instruction
  - path: skills/code-review/SKILL.md
    kind: skill

Validation Rules:

id: lowercase, hyphens/numbers only
name: 1-100 characters
description: 1-500 characters
items: 1-50 items, paths must exist
kind: prompt, instruction, chat-mode, agent, or skill

MCP Servers (Optional)

Add Model Context Protocol servers to provide custom tools:

# In collections/*.collection.yml
mcp:
  items:
    # Pre-built server (recommended)
    time:
      command: npx
      args:
        - -y
        - "@modelcontextprotocol/server-sequential-thinking"
    
    # Custom server (advanced)
    custom:
      command: node
      args:
        - ${bundlePath}/mcp-server/server.js

See mcp-server/README.md for available servers and custom implementation guides.

🧪 Testing Workflow

Local Validation

# Run all validation
npm run validate
npm run skill:validate

# Or with Node directly
node validate-collections.js
node scripts/validate-skills.js

CI/CD (GitHub Actions)

The included workflow (.github/workflows/validate-collections.yml) runs automatically:

✅ On every push to main/develop
✅ On pull requests
✅ Reports validation errors
✅ Blocks merge if validation fails

Manual Testing

In Copilot Chat: Use / to access prompts
With AI Primitives Hub: Browse and install collections
Validate Files: Check YAML syntax and file references
Validate Skills: Verify SKILL.md frontmatter
MCP Servers (if enabled): Verify server appears in VS Code MCP settings

📋 Quality Checklist

Before committing:

npm install completed successfully
npm run validate passes with no errors
npm run skill:validate passes with no errors
File naming follows conventions
All collection paths exist
YAML syntax is valid
VS Code shows no schema errors
MCP configuration tested (if enabled)

📚 Resources

🛠️ Extension Commands

Available when using AI Primitives Hub extension:

AI Primitives Hub: Validate Collections - Validate all collections
AI Primitives Hub: Create New Collection - Interactive collection wizard
AI Primitives Hub: List All Collections - View collection metadata
AI Primitives Hub: Add Resource - Add prompt/instruction/agent/skill

📄 License

Apache License 2.0 - see LICENSE

🙏 Acknowledgments

Based on github/awesome-copilot structure and best practices.

Next Steps: Review examples → Run npm install → Run npm run validate → Run npm run skill:validate → Create your first collection! 🚀

Name		Name	Last commit message	Last commit date
.githooks		.githooks
.github		.github
collections		collections
foursight		foursight
ospo		ospo
skill-eval		skill-eval
.gitignore		.gitignore
CODE_OF_CONDUCT.md		CODE_OF_CONDUCT.md
COMMUNICATION.md		COMMUNICATION.md
CONTRIBUTING.md		CONTRIBUTING.md
LICENSE		LICENSE
README.md		README.md
SECURITY.md		SECURITY.md
package-lock.json		package-lock.json
package.json		package.json

Uh oh!

Latest commit

History

Folders and files

Repository files navigation