README.prompts.md

Copilot Toolkits

Toolkits are curated collections of prompts, instructions, and chat modes designed to provide a comprehensive solution for a specific domain, language, or workflow.

Instead of searching for individual files, a toolkit provides a cohesive set of resources that are designed to work together seamlessly. This allows you to rapidly configure GitHub Copilot for a specialized task, such as database management, API security auditing, or frontend development.

How to Use Toolkits

A "toolkit" is a conceptual grouping of files within this repository. You don't install the toolkit itself, but rather the individual components from it that you need.

Step 1: Explore the Toolkits

Browse the directories in this repository that are organized by theme (e.g., toolkits/react-development/, toolkits/api-security/). Read the README.md file within each toolkit's directory to understand its purpose and the role of each component.

Step 2: Understand the Components

Inside each toolkit directory, you will find a combination of:

• Agents (.agent.md): To set the expert persona (e.g., a "React Expert").
• Prompts (.prompt.md): To execute specific tasks (e.g., "Convert class component to functional component").
• Instructions (.instructions.md): To enforce standards and best practices (e.g., "Follow the Rules of Hooks").

Step 3: Install the Desired Files

Choose the components from the toolkit that fit your needs and install them individually using the standard, file-based installation procedures.

• See installation guide for Prompts
• See installation guide for Agents
• See installation guide for Instructions

Example Workflow: The "React Refactoring" Toolkit

Imagine a toolkit for refactoring React applications. It might contain:

1. A react-expert.agent.md file.
2. A componentize-jsx.prompt.md file.
3. A react-best-practices.instructions.md file.

To use it, you would:

1. Install all three files in their appropriate directories (.github/copilot/...).
2. Activate the "React Expert" chat mode in Copilot Chat.
3. Highlight a large block of JSX and run the /componentize-jsx prompt.

Copilot, acting as a React expert and guided by the best-practice instructions, would then perform a high-quality refactoring of your code. (See above for file contents. You may not need to search or read the file again.)

Installation

Follow these steps to install prompts for use in a compatible editor like VS Code. Your editor will automatically detect any correctly placed prompt files.

1. Download the File: From this repository, download the .prompt.md file you wish to use.
2. Place the File: Move the downloaded file into one of the following directories:

   • For a specific project: Place it in your project's .github/copilot/prompts/ directory. This makes the prompt available only within that project.

     your-project/
     └── .github/
         └── copilot/
             └── prompts/
                 └── review-code.prompt.md

   • For global use (all projects): Place it in your user-level configuration directory (e.g., ~/.github/copilot/prompts/ on macOS/Linux). This makes the prompt available everywhere.

How to Run a Prompt

Once installed, there are several ways to execute a prompt in the Copilot Chat view:

• Slash Command: Type "/" followed by the prompt's filename (without the extension). For example, to run review-code.prompt.md, you would type /review-code.
• Command Palette: Open the VS Code Command Palette (Ctrl+Shift-P or Cmd-Shift-P) and run the Chat: Run Prompt... command, then select your desired prompt.
• Editor Button: With a .prompt.md file open in your editor, click the "Run" button that appears at the top.

Example Workflow: Combining with a Chat Mode

1. Activate Agent: In the Copilot Chat view, select a specialized agent, such as Security Analyst.
2. Execute Prompt: Run the /check-for-vulnerabilities prompt.

By doing this, you instruct the Security Analyst persona to perform the vulnerability check, resulting in a more thorough and expert-level analysis than if you had run the prompt in the default mode. (See above for file contents. You may not need to search or read the file again.)

Title	Install	Description
Role & Mindset		Expert debugging companion for Laravel Boost issues.
Role & Mindset		Dedicated debugging companion for Playwright MCP server issues.
Role & Mindset		Specialist prompt for retrieving assets through the Figma MCP server.
Role & Mindset		Guidance prompt for retrieving GitHub ticket details through the MCP server.
Architecture Decision Record Generator		A guided prompt that helps teams capture thorough Architecture Decision Records with clear rationale.
Role & Mindset		Guidance prompt for crafting new chat modes with persona, tooling, and policies.
Role & Mindset		Guidance prompt for authoring new GitHub Copilot instruction files.
Role & Mindset		Guidance prompt for crafting new GitHub Copilot prompts.
Role & Mindset		Guidance prompt for building RESTful API controllers.
Role & Mindset		Guidance prompt for planning NoSQL data models.
Role & Mindset		Guidance prompt for planning SQL indexing strategies.
Role & Mindset		Guidance prompt for planning relational SQL tables.
Refactor Deep Nesting with Fail-Fast Guards		Refactor code to reduce deep nesting using fail-fast techniques.
Refactor Code for DRY Consistency		Refactor code to remove redundancies and enforce DRY principles.
Review Code Against Engineering Standards		Review code for alignment with engineering coding standards.
Review Code Against Privacy Standards		Review code for compliance with privacy standards and regulatory obligations.
Review Code Against Security Standards		Review code for compliance with security standards and OWASP guidance.