If you've ever wanted an API client that plays nicely with Git, you've probably discovered Bruno. But did you know you can make Bruno even more powerful by teaching GitHub Copilot to understand its collection schema? Let me show you how.
Bruno uses plain-text markup language to store API requests. While this is perfect for version control and team collaboration, AI coding assistants like GitHub Copilot don't understand this format out of the box. Without guidance, Copilot might suggest JSON-based request formats (like Postman uses) or generic HTTP client code instead of proper syntax.
GitHub Copilot supports custom instruction files that teach it about your project's specific patterns and conventions. By adding a simple markdown file to your Bruno project, you can transform Copilot from a generic coding assistant into a Bruno-aware powerhouse.
The fastest way to get started is with our automated installer:
cd your-bruno-project
curl -fsSL https://raw.githubusercontent.com/bruno-collections/ai-assistant-prompts/main/install.sh | bash
Or manually create .github/copilot-instructions.md in your project root and copy the Bruno Copilot instructions.
That's it! Restart your editor and Copilot will now understand Bruno.
Without instructions, Copilot might suggest something like this when you create a new API request:
// Copilot suggests generic fetch code
const response = await fetch('https://api.example.com/users', {
method: 'POST',
headers: { 'Content-Type': 'application/json' },
body: JSON.stringify({ name: 'John' })
});
With the instruction file in place, Copilot suggests a full request with proper syntax generated just from the request name itself.
First, we'll clone FastAPIs repo, https://github.com/FastAPI/FastAPI, and add .github/copilot-instructions.md with with the Bruno instructions.
Now I'll just ask Copilot to generate the collection:
Just like that, Copilot has created a `bruno` folder inside the codebase with a full collection, organized nicely into folders based on category.
Copilot has also stubbed out environments for us as well:
Now let's open this in Bruno and experience the collection!
The full collection is now available in Bruno! I also asked Copilot to create a simple test server so we're able to hit the endpoints in the collection.
I've intentionally written a script with incorrect syntax that's resulting in a failure:
I'll just ask Copilot to fix it 😀
Beautifully fixed.
We'll use the collection runner to run all the requests and tests at once:
When I ran the collection, you'll notice some failures like mismatched prices or IDs.
These failures demonstrate that Copilot generated real-world test assertions, not only basic "does it return 200 OK?" checks:
The key insight: Copilot didn't just generate happy-path tests. It created assertions that validate business logic, data integrity, and error handling, the same tests you might write by hand for a production API.
When your entire team has the same Copilot instructions, everyone gets consistent suggestions. This means:
Since the instruction file is just a markdown file in your repository, it:
Instead of request1.bru, use create-user.bru or get-user-profile.bru. Copilot uses file names as context.
Start your .bru files with comments explaining the purpose:
# Creates a new user account with email verification
# Requires: authToken (admin), userName, userEmail
# Returns: userId, verificationToken
meta {
name: Create User with Verification
type: http
seq: 1
}
Let Copilot help you build requests step by step:
meta blockAt each step, Copilot will suggest the next logical section.
Keep a few well-structured .bru files open in your editor. Copilot uses open files as additional context.
The bruno-ai-assistant-prompts repository includes instruction files for multiple AI assistants:
.cursorrules file for Cursor editorAll of these teach AI assistants the same Bruno patterns, so you can use whichever tool fits your workflow.
Bruno's Git-first, offline-only approach is perfect for modern API development, and with proper Copilot instructions, you can make your AI assistant understand Bruno as well as you do. The setup takes less than a minute, but the productivity gains compound over time.
Whether you're working solo or with a team, teaching Copilot about Bruno means faster