Best Practices

Master the art of working with Copilot Chat. Learn how to write effective prompts, manage context, handle privacy concerns, integrate into team workflows, and avoid common pitfalls.

Writing Effective Chat Prompts

The quality of Copilot Chat's response depends heavily on how you phrase your prompt. Follow these principles for better results:

Be Specific and Detailed

// Vague prompt (poor results)
Write a function to process data

// Specific prompt (much better results)
Write a TypeScript function called processUserData
that takes an array of User objects, filters out
inactive users (isActive === false), sorts the
remaining by lastLoginDate descending, and returns
the top 10 with only id, name, and email fields

Provide Context

// Include constraints and requirements
Write a caching middleware for Express that:
- Caches GET responses in Redis
- Uses the request URL as the cache key
- Expires after 5 minutes
- Skips caching for authenticated routes
- Returns appropriate Cache-Control headers
- Uses the existing Redis client from #file:src/config/redis.ts

Break Down Complex Tasks

// Instead of one huge prompt, break it down:

// Step 1: Design
What is the best way to implement rate limiting
for our Express API? Consider our architecture in
@workspace

// Step 2: Implement
Implement the sliding window rate limiter you
described using Redis

// Step 3: Test
/tests Generate tests for the rate limiter including
edge cases for concurrent requests

// Step 4: Document
/doc Add comprehensive documentation

When to Use Chat vs Inline vs Quick Chat

Choosing the right interaction mode makes your workflow more efficient:

Context Management Best Practices

Effective context management is the single biggest factor in getting good results from Copilot Chat.

• Start fresh for new topics: Use /clear when switching to a completely different task. Stale context from previous conversations can confuse responses.
• Select relevant code: Before asking a question, select the specific code you are asking about. Do not rely on Copilot to guess which part of the file you mean.
• Use #file for cross-file context: When your question involves code in another file, explicitly reference it with #file. Do not assume Copilot knows about other files.
• Use @workspace sparingly: @workspace searches your entire project, which adds latency. Use it for project-wide questions, not for questions about specific code.
• Keep conversations focused: One topic per conversation. If you need to switch topics, start a new chat.

Privacy and Data Handling

Understanding how Copilot Chat handles your code is important, especially in corporate environments.

Team Workflow Integration

Teams get the most value from Copilot Chat when they establish shared practices:

1. Establish Custom Instructions

   Create a shared .github/copilot-instructions.md that encodes your team's coding standards, architecture decisions, and testing practices.

2. Create Shared Prompt Templates

   Build a library of custom slash commands in .github/copilot-prompts/ for common team tasks like code review, migration patterns, and documentation standards.

3. Standardize on Workflows

   Agree on when to use chat vs inline vs edits mode. Document recommended prompts for common scenarios in your team wiki.

4. Review AI-Generated Code

   Treat all AI-generated code as you would any developer's code: it must go through code review, pass tests, and meet quality standards.

Common Mistakes to Avoid

Frequently Asked Questions