Copilot Toolkit Banner

Copilot Toolkit — AI Code Review & Prompt Automation

Supercharge GitHub Copilot with smart skill-driven code review, multi-step workflow automation, dynamic prompt libraries, auto skill detection, and adaptive AI recommendations.

Overview

Copilot Toolkit is a VS Code extension that adds a powerful orchestration layer on top of GitHub Copilot Chat. Instead of typing raw questions, you compose skill instruction sets (your engineering standards) with reusable prompts (your review tasks) and send them together as structured, high-quality requests.

It works with any language, framework, or stack — the extension reads your project and open file to automatically recommend the right skills, learns your preferences over time, and lets you build multi-step AI workflows with just a few clicks.

Features

🧠 Smart Skill System

Define your engineering standards as Markdown files. The extension composes them into a rich instruction context sent to Copilot before every request.

• Default skill from .github/copilot-instructions.md (respects your existing Copilot instructions)
• Custom skill sets from .copilot/skills/*.md (security, performance, frontend, backend, etc.)
• Multi-skill merge — select multiple skills; each is labelled and combined with a priority header

⚡ Auto Skill Detection

The extension scans your project and active file to detect:

🎯 Adaptive Learning

Copilot Toolkit remembers which skills you choose and learns your patterns:

• Tracks individual skill usage counts
• Detects frequently used skill combinations
• Surfaces a "You frequently use: X + Y" hint in the recommendation UI
• Boosts frequently used skills to the top of the picker

Reset anytime — run Copilot Toolkit: Reset Learned Preferences from the Command Palette.

📂 Dynamic Prompt Library

5 built-in prompts are always available, plus unlimited user-defined prompts:

Add your own by dropping .md files into .copilot/prompts/.

🔄 Multi-Step Workflow Mode

Select multiple prompts and run them as a single sequential request to Copilot Chat. Perfect for thorough reviews that cover multiple angles at once (e.g. Debug + Performance + Security in one shot).

👁 Skill Preview

Before sending your request, open a read-only Markdown tab to inspect the combined skill instructions — so you always know exactly what context Copilot is receiving.

Installation

From the Marketplace

1. Open VS Code
2. Press Ctrl+Shift+X (Extensions panel)
3. Search for Copilot Toolkit
4. Click Install

From VSIX (manual)

code --install-extension copilot-toolkit-1.0.0.vsix

Prerequisite: GitHub Copilot must be installed and active in your VS Code.

Usage

Quick Start

1. Open any file in your project
2. Press Ctrl+Shift+P → Copilot Toolkit
   or right-click in the editor → Copilot Toolkit
3. Choose a mode: Single Prompt or Multi-Step Workflow
4. Review the smart skill recommendation (or customize manually)
5. Pick a prompt
6. Copilot Chat opens with your structured request

Keyboard Shortcut (optional)

Add to your keybindings.json:

{
  "key": "ctrl+shift+alt+c",
  "command": "copilot-toolkit.open"
}

Skills Setup

Default Skill

Create .github/copilot-instructions.md in your project root. This file is automatically used as the default skill and also feeds GitHub Copilot's native instruction system.

# My Engineering Standards

- Always prefer immutable data structures
- All async functions must handle errors explicitly
- Prefer composition over inheritance
- All public APIs must have JSDoc comments

Additional Skills

Create .copilot/skills/ in your project root and add .md files:

.copilot/
└── skills/
    ├── security.md
    ├── performance.md
    ├── frontend.md
    └── database.md

Example: security.md

# Security Review Standards

- Validate and sanitize all user inputs before processing
- Never log sensitive data (passwords, tokens, PII)
- Use parameterized queries — never string-concatenated SQL
- All auth tokens must have expiry and be rotated after use
- Check OWASP Top 10 patterns: injection, XSS, CSRF, broken auth

Prompts Setup

Create .copilot/prompts/ and add .md files for custom prompts:

.copilot/
└── prompts/
    ├── api-review.md
    ├── accessibility.md
    └── migration-check.md

Example: api-review.md

Review this API endpoint based on the provided engineering instructions.
Check: input validation, authentication, authorization, rate limiting,
error response format, HTTP status codes, and logging.

Configuration

All paths are relative to the workspace root.

Example settings.json:

{
  "copilotToolkit.skillsFolder": ".ai/skills",
  "copilotToolkit.promptFolder": ".ai/prompts",
  "copilotToolkit.enableLearning": true
}

Folder Structure

your-project/
├── .github/
│   └── copilot-instructions.md    ← default skill
└── .copilot/
    ├── skills/
    │   ├── security.md
    │   ├── performance.md
    │   └── frontend.md
    └── prompts/
        ├── api-review.md
        └── migration-check.md

Commands

How It Works

You trigger Copilot Toolkit
        ↓
Auto-detect project tech stack + active code patterns
        ↓
Smart recommendation: "Use Recommended Skills" or "Customize Manually"
        ↓
Pick prompt(s) — single or multi-step workflow
        ↓
Build structured prompt: [Header] + [Skills] + [Task] + [Code]
        ↓
Open GitHub Copilot Chat with the full prompt
        ↓
Adaptive learning records your skill choices

Requirements

• VS Code ^1.90.0
• GitHub Copilot extension (installed and authenticated)

License

MIT © 2024 Copilot Toolkit Contributors

Contributing

Issues, feature requests, and PRs are welcome at the GitHub repository.