A community-driven registry for Claude, Cursor, Windsurf, Cline & more. Not affiliated with Anthropic.
Are you the author? Sign in to claim
A curated collection of GitHub Copilot prompts, instructions, chat modes, and MCP servers configurations for modern web
A curated collection of GitHub Copilot prompts, instructions, and chat modes for modern web development
This repository contains battle-tested GitHub Copilot configurations that I use daily and showcase on my YouTube channel. Each prompt and configuration is designed to boost your productivity and help you write better code faster.
Star ⭐ this repo to keep up with new additions and updates!
Fine-tuned instructions for specific frameworks and patterns
| Instruction | Description | Tech Stack |
|---|---|---|
| Coming soon | Custom instructions for specific frameworks | Planned |
Ready-to-use prompts for common development tasks
📖 Read the Custom Prompts Guide for detailed documentation on how to use and create prompts.
Prompts are starting points, not one-shot solutions—this is AI-assisted programming. Run the prompt, optionally provide additional context, then iterate on the results until you're satisfied.
flowchart LR
A[Run Prompt] --> B[Review Results]
B --> C{Satisfied?}
C -->|No| D[Iterate: Ask/Answer/Clarify]
D --> B
C -->|Yes| E[Done ✓]
New to this repo? Start with the public prompt entrypoint:
Recommended default: /kit-customize-copilot
Run the prompt in Copilot Chat and describe the customization outcome you want. For example:
"/kit-customize-copilot Create project-wide Copilot guidance for this repository based on its tech stack and conventions"
This routes to the kit-copilot-engineering-team orchestrator, which inspects the repo, classifies the request, and chooses the right internal workflow for you.
If you prefer to go deeper after the first pass, read the dedicated guide: Customize Copilot with /kit-customize-copilot.
Optional direct path: if you like working from the agent picker, you can select kit-copilot-engineering-team directly and describe the same outcome there.
The repository-owned kit-copilot-create-* skills still exist as internal building blocks, but they are no longer the default onboarding path or the recommended public alternative to /kit-customize-copilot.
Why does this matter? Without custom instructions, Copilot relies on generic knowledge. With them, it understands your project's conventions, patterns, and tech stack — producing more consistent, relevant, and accurate suggestions across your entire codebase.
Use these prompts for day-to-day development tasks.
| Prompt | Description | Category |
|---|---|---|
| architecture-review | Focused architectural review using Sequential Thinking for multi-dimensional analysis | Architecture |
| code-review | High-rigor, tech-agnostic code review that surfaces critical issues first | Code Quality |
| pragmatic-code-review | Production-focused code review balancing engineering rigor with real-world constraints | Code Quality |
| refactor | Intelligent refactoring following user instructions or applying best practices automatically | Code Quality |
| optimize-performance | Identify and eliminate performance anti-patterns using static analysis and language-idiomatic optimizations | Performance |
| document | Generate exceptional documentation with Mermaid diagrams and practical examples | Documentation |
Specialized Copilot agent personalities for different development scenarios
| Agent | Description | MCP Tools |
|---|---|---|
| kit-copilot-engineering-team | Copilot engineering orchestrator — handles the full range of customization requests from single-mechanism tasks to complex multi-artifact projects. Delegates to specialized workers for research, creation, validation, and review. | Sequential Thinking |
| kit-backend-engineer | Pragmatic backend engineer specializing in APIs, databases, and server-side architecture | Sequential Thinking, Context7 |
.github/agents/ directory (VS Code automatically detects .agent.md files in this folder).vscode/mcp.json (see MCP Servers section)Note: Custom agents require VS Code 1.106 or later.
The kit-copilot-engineering-team agent is the optional direct-agent path for Copilot customization in this repository. If you prefer using the agent picker instead of the public prompt, select it directly and describe what you want to achieve.
For most users, the recommended public starting point remains /kit-customize-copilot.
Example prompts:
The orchestrator delegates to specialized workers: the researcher discovers what exists and classifies what's needed, the creator builds the artifacts, the validator runs checks, and the reviewer ensures quality. For simple unambiguous requests, it uses a lightweight create-and-review path.
Want to understand not just what is in this repository, but why it is structured this way? Browse the ADRs for the architectural and workflow decisions behind copilot-kit.
More ADRs will be added over time as the project evolves.
copilot-kit uses one primary orchestrator with curated domain orchestratorsReusable skill artifacts that extend Copilot's capabilities. Skills are defined as SKILL.md files following the Agent Skills specification, VS Code Skills docs, and can be adapted for your own repositories.
For the Copilot customization workflow in this repository, the preferred public slash-command entry is /kit-customize-copilot. The repository-owned kit-copilot-create-* skills below are documented as reusable building blocks and reference implementations — not as the default public onboarding path.
| Skill | Description |
|---|---|
| kit-copilot-create-agent | Scaffolds a custom agent (.agent.md) file for GitHub Copilot with persona, philosophy, tools, and workflow. Useful as a reusable building block when creating a new custom agent, AI persona, or chat mode. |
| kit-copilot-create-commit-instructions | Generates a commit message instruction file for Copilot's commit message generation in VS Code. Useful for standardizing commit-message behavior in a repo. |
| kit-copilot-create-core-instruction | Creates the foundational .github/copilot-instructions.md file that provides universal project context for Copilot interactions. |
| kit-copilot-create-instruction | Creates fine-grained .github/instructions/*.instructions.md files with targeted applyTo glob patterns for specific technologies, file types, or architectural layers. |
| kit-copilot-create-prompt | Creates a simple Copilot prompt file (.prompt.md) for lightweight, repeatable workflows. |
| kit-copilot-create-skill | Scaffolds an Agent Skill directory with SKILL.md and optional bundled resources for richer reusable workflows. |
If you want to study, copy, or adapt Agent Skills for your own repository, there are two common ways to make them available. For this repository's Copilot customization workflow, though, start with /kit-customize-copilot instead of manually choosing among the repository-owned kit-copilot-create-* skills.
Copy the skill directory into your project's .github/skills/ folder. This lets you adjust the skill to fit your project's specific needs.
Point VS Code to this repo's skills directory so reusable skills are available across projects — no copying needed.
Cmd + , on macOS, Ctrl + , on Windows/Linux)chat.agentSkillsLocations)~/dev/ikcode-dev/copilot-kit/.github/skills
Tip: If you clone this repo to a different path, update the setting accordingly. Use
~for your home directory to keep it portable across machines.
What is MCP?
.vscode/mcp.json plus a memory template (it's empty, MCP will fill it out) at .mcp/memory.json.dist..vscode/mcp.json - a sample configuration wiring up the servers below.mcp/memory.json.dist - template for local, per-project memory storagePrerequisites
Copy the memory storage template (if you want to use memory MCP) .mcp/memory.json.dist and save it as .mcp/memory.json
memory.json is an empty file, so you can just create empty file at .mcp/memory.json.mcp/memory.json.dist to the Git repo if you work with a team.mcp/memory.json file, it'll contain data relevant to your workCopy configurations of the servers you want to use from the .vscode/mcp.json to your project
${input:input-id}in the env or other section, this means it's going to store some env variables in VSCode and the specific input is defined in the inputs field of the JSON configuration object at .vscode/mcp.json.input is required for a MCP server (like for shadcn-ui) you will be prompted for the value on the server's first startup, then it's going to be saved securely by VSCode.inputs is my preferred way of storing env variables for MCP servers. I want to commit .vscode/mcp.json, so my team members can use them too anytime they want, without any danger of sharing the real env variables.| Name | What it gives you | Notes |
|---|---|---|
| Sequential Thinking | This MCP gives the agent ability use more structured thinking process, it's really powerful when combined with the right prompts | 🌟 I use it everyday, basically all the time |
| Memory | It gives your agent ability to memorise different things between conversations, it's really powerful when connected with other MCP servers or remembering some things about the project | ♥️ I combine it with Context7 to remember libraries IDs |
| Context7 | It allows your agent to have the best context when it comes to the newest versions of tools, libraries, frameworks, etc. | 🆓 Free for all developers! I use this whenever I need up-to-date library context. Highly recommended for every project. |
| Shadcn UI | It gives the agent access to the Shadcn UI components and helps understand how they should be used | To me it's absolute banger. 🚀 I use it whenever I work with Shadcn UI in my projects. |
| Chrome DevTools | Brings Chrome DevTools debugging capabilities to your agent - inspect live pages, debug network/console errors, simulate user behavior, analyze performance, and verify code changes in real-time | 🔍 Perfect for debugging web apps and performance audits |
All things in this repo are optimized for:
Found a prompt that supercharged your workflow? Feel free to contribute!
If you have a question or want to talk about some idea related to this repo - feel free to open an issue.
MIT License - Use these prompts freely in your projects!
⭐ If you find this useful, please star the repository! It helps others discover these resources and motivates me to add more content.
Run Claude Code as an MCP server so any agent can delegate coding tasks to it
@microsoft✓ OfficialBrowser automation using accessibility snapshots instead of screenshots
MCP server integration for DaVinci Resolve Studio
A Jetbrains IDE IntelliJ plugin aimed to provide coding agents the ability to leverage intelliJ's indexing of the codeba