GitHub Copilot

GitHub Copilot is an enterprise AI pair programming assistant, providing inline completions, multi-file code editing via Copilot Edits, pull request reviews, and CLI integration.

When to Use

• In-Editor AI Pair Programming: Generating functions, documentation, and boilerplate across VS Code, JetBrains, and Neovim.
• Repository-Level Architectural Context: Using Copilot Chat with workspace awareness (@workspace).
• Standardizing Team Prompting: Authoring .github/copilot-instructions.md to enforce organizational coding standards.
• Automated Pull Request Summarization: Generating detailed PR summaries and test plans within GitHub Enterprise.

Quick Start

Configure custom repository instructions in .github/copilot-instructions.md:

# GitHub Copilot Instructions

- Use modern ES2022+ syntax and TypeScript 5.
- Always validate incoming API payloads using Zod schemas.
- Do not add comments describing self-explanatory lines.
- Write unit tests using Vitest with AAA (Arrange-Act-Assert) pattern.

Core Concepts

Project-Wide Custom Instructions (.github/copilot-instructions.md)

Configuring persistent AI developer guidelines:

# .github/copilot-instructions.md

Repository Standards

- Language: TypeScript 5.5+ with strict mode enabled.
- Framework: Next.js 15 (App Router), Tailwind CSS v4.
- Database: Drizzle ORM with PostgreSQL.
- Testing: Vitest for unit tests, Playwright for e2e.

Coding Conventions

1. Always declare explicit return types on public functions.
2. Prefer async/await over raw Promises.
3. Use early returns and guard clauses to minimize nested blocks.
4. Never suggest 'any'; use typed generics or unknown.
5. All new business functions must include an accompanying unit test file.

Slash Commands & Context Modifiers in Copilot Chat

Directing the AI assistant with precision:

# Prompt in Copilot Chat
@workspace /explain How is user authentication state persisted across server and client components?

# Generating Unit Tests
/tests Generate comprehensive unit tests covering edge cases for validateOrderPayload()

# Fixing Errors
@terminal /fix Resolve this TypeScript compilation error with discriminated unions

Inline Prompt Engineering for Precise Code Generation

Guiding inline suggestions with intentional comment prompts:

// Function: parseJWTHeader
// Inputs: raw auth header string (e.g., 'Bearer <token>')
// Output: decoded payload object or throws UnauthorizedException
// Handles expired tokens, malformed base64, and missing prefix
export function parseJWTHeader(authHeader?: string): TokenPayload {
  // Copilot autocompletes implementation here
}

Common Patterns

Custom Copilot Instructions (.github/copilot-instructions.md)

Problem: Generic completions ignore repository style guides, testing patterns, and library versions.
Solution: Create project-level guidelines recognized by Copilot in VS Code and JetBrains.

<!-- .github/copilot-instructions.md -->

# Project Guidelines

- Architecture: Feature-sliced design (`src/features/<feature-name>/{api,components,hooks}`).
- State Management: TanStack Query for server state; Zustand for global client state.
- Testing: Write Vitest unit tests colocated as `<name>.test.ts` for all pure helper functions.
- Concurrency: Prefer `Promise.allSettled()` over `Promise.all()` for batch operations.

Prompt-Driven Unit Test Generation (/tests)

Problem: Writing repetitive unit test boilerplate for edge cases and validation schemas.
Solution: Highlight target function in editor and invoke /tests with constraints.

// Prompt: /tests generate vitest test cases covering boundary values, empty arrays, and null inputs
export function calculateDiscount(price: number, coupon?: string): number {
  if (price < 0) throw new Error("Price cannot be negative");
  if (coupon === "VIP50") return price * 0.5;
  if (coupon === "SAVE10") return Math.max(0, price - 10);
  return price;
}

Best Practices

Do:

• Maintain a .github/copilot-instructions.md in repository root to align Copilot with team conventions.
• Use @workspace to scope prompts to whole repository architecture and dependency patterns.
• Provide clear docstrings, parameter types, and test names to guide high-quality completions.
• Verify and write unit tests for all Copilot-generated logic before committing.

Don't:

• Accept multi-line completions without verifying security implications (e.g. SQL injection, sanitization).
• Prompt Copilot with confidential tokens, production certificates, or customer PII.
• Rely on Copilot for sensitive cryptographic algorithm implementation without formal review.

Troubleshooting

References

• GitHub Copilot