A collection of custom instructions and prompts to enhance your GitHub Copilot experience.
This repository contains curated .instructions.md and .prompt.md files that you can use to customize GitHub Copilot's behavior across your projects. These files help Copilot understand your preferred workflows, coding patterns, and best practices.
Instructions files (.instructions.md) define rules and guidelines that Copilot follows when generating or modifying code. They can be applied globally or to specific file patterns.
-
azure-devops-mcp.instructions.md-- Guidelines for working with Azure DevOps MCP Server
- Work item creation and management best practices
- Naming conventions and field requirements
- Linking and relationship management
-
- Conventional Commits specification guidelines
- Commit type definitions (feat, fix, docs, etc.)
- Scope and description formatting rules
- Issue reference recommendations
Prompt files (.prompt.md) are reusable task templates that guide Copilot through complex multi-step operations.
-
- Creates pull requests on Azure DevOps
- Auto-extracts work item numbers from commits
- Generates PR title and description from commit history
- Supports configurable target branch
-
init-github-repository.prompt.md-- Automates GitHub repository setup
- Analyzes workspace and creates appropriate repository
- Initializes git and pushes code
- Handles authentication and error scenarios
-
- Generalizes the current discussion into a reusable prompt
- Saves the generated prompt to a file
Agent files (.agent.md) define specialized AI agents with specific expertise and capabilities for focused tasks.
-
custom-agent-creator.agent.md-- Specialized agent for creating custom Copilot agents
- Guides through agent design and configuration
- Helps define agent capabilities and behavior
-
microsoft-learn-doc-scout.agent.md-- Doc-first research agent using Microsoft Learn MCP server
- Answers questions with cited Microsoft documentation
- Covers Azure, .NET, and M365 workloads
- Provides actionable summaries and guidance
-
website-retrospec-generator.agent.md-- Generates retrospective analysis for websites
- Analyzes web architecture and design patterns
- Provides improvement recommendations
Click the Install in VS Code or Install in VS Code Insiders button next to any instruction, prompt, or agent to install it directly.
To Install:
- Click the install button above for one-click installation, or
- Download the
*.instructions.md,*.prompt.md, or*.agent.mdfiles and manually add them to your VS Code settings
To Use/Apply:
- Instructions automatically apply to Copilot behavior once installed
- Prompts can be referenced in Copilot Chat by name
- Agents become available as specialized participants in Copilot Chat
You can also add these files to individual workspaces by placing them in your workspace's .github/instructions folder.
Once installed, instructions are automatically applied based on their applyTo frontmatter:
---
description: 'Guidelines for using Azure DevOps MCP Server'
applyTo: '**'
---**applies to all files in the workspace- Specific patterns like
*.tsapply only to TypeScript files
Prompts can be invoked in Copilot Chat:
- Open GitHub Copilot Chat (
Ctrl+Shift+IorCmd+Shift+I) - Reference the prompt by name or use the prompt picker
- Copilot will follow the steps defined in the prompt
Example:
@workspace Use the init-github-repository prompt to set up this project
Once installed, custom agents (chat modes) can be selected in the Chat view:
- Open GitHub Copilot Chat (
Ctrl+Shift+IorCmd+Shift+I) - Select the desired agent from the chat mode dropdown list
- The agent will respond with its specialized knowledge and capabilities according to its configuration
Custom agents define specialized personas and available tools for specific development tasks like planning, code review, or implementation.
Feel free to use, modify, and share these instructions and prompts. Contributions are welcome!
Created to enhance developer productivity with GitHub Copilot.
Note: These instructions and prompts work best with GitHub Copilot Chat in VS Code. Ensure you have the latest version of the GitHub Copilot extension installed.