News

GitHub Copilot Is Rising All-Time Contributor to .NET MAUI Repo

• By David Ramel
• 07/15/2025

Microsoft's .NET MAUI dev team is putting GitHub Copilot to good use with DevOps tasks that make it a rising all-time code contributor to the repo, along with all the other AI assistance it provides.

"At the time of writing, Copilot has risen to be the #64 contributor all time in dotnet/maui, which we expect to grow in the coming months," the team's Jonathan Peppers, principal software engineer, said in a July 9 DevBlogs post highlighting the tools' internal productivity.

That's en route to the team's goal of realizing a "dream scenario" of:

• A GitHub issue is filed.
• The problem looks to have a straightforward solution.
• A team member assigns the issue to Copilot, giving it a few tips.
• Copilot generates a pull request with the fix (and tests).
• CI runs, tests pass, and the pull request is merged!

In testing this out, the team found it to sometimes be "genuinely helpful" while at other times "it completely missed the mark." In other words, in our editorial opinion, it's pretty much like just about every other AI assistant out there that we've seen.

Tips for Putting Copilot Agent to Work
To move closer to the automated ideal, the .NET MAUI team outlined a structured approach. Step one is providing Copilot with proper context. "To start with the basics, we provide GitHub Copilot with some general context and guidance for our repository by including a copilot-instructions.md file," the team explained. "By convention this file should be kept in the .github folder at the root of your repository or workspace."

That file includes information about project context, repository structure, and any custom coding standards. "A great way to start is to simply ask the Copilot Coding Agent to generate this file for you," the post suggested.

Next is configuring firewall rules to control outbound access. When Copilot tries to pull external resources during a task, GitHub will block unauthorized domains. To avoid failures, users can set a COPILOT_AGENT_FIREWALL_ALLOW_LIST environment variable that specifies permitted domains like learn.microsoft.com and nuget.org. "This warning is actually a key security feature," the post noted, referring to the firewall block message.

A third component is the setup workflow. The team uses a GitHub Actions workflow defined in .github/workflows/copilot-setup-steps.yml to initialize the build environment before Copilot begins work. The steps include restoring packages, performing a baseline build, and preparing the project structure for incremental changes. The team advised making this workflow fault-tolerant by setting continue-on-error: true: "Copilot (or a human) might push a commit that breaks the build," they wrote. "If you leave a comment like @copilot fix error XYZ, it needs to be able get past its setup steps and actually fix the problem."

Finally, the .NET MAUI team optionally integrates a Model Context Protocol (MCP) server to give Copilot access to external tools. "In our experience, the Microsoft Learn Docs MCP Server is one of the most powerful tools available," the blog noted. This lets Copilot reference Microsoft documentation during tasks such as code modifications.

Real-World Example: CatSwipe
In one experiment, the team asked Copilot to take a screenshot of a sample .NET MAUI app. Instead, it:

• Wrote a C# console app.
• Used System.Drawing to generate an image of an Android emulator and what it imagined the app would look like.
• Checked in the screenshot, "as if it was a real screenshot!"

As a result, they adjusted their strategy. "When assigning issues, always add links to documentation. Write the issue in a way that a junior engineer (human) could pick up the task," they advised. "When Copilot does the wrong thing, it likely needs more context or more instructions. Think of this as 'debugging' your copilot-instructions.md file."

Early Results and Limitations
The .NET MAUI team also shared initial usage metrics. In the dotnet/android repo, Copilot authored 17 pull requests with an 82.4% merge rate and a median merge time of 10:15:34. By comparison, all other contributors had a slightly higher merge rate (87.8%) but slower median time (11:36:35).

In the dotnet/maui repo, where the team intentionally assigned harder issues, Copilot authored 54 PRs with a 16.7% merge rate versus 52.9% for others. "We've been more optimistic: assigning PRs we knew might be too tough for Copilot to complete," they explained. "Take these numbers with a grain of salt, as we have certainly been focusing a decent amount of time on Copilot."

Current limitations include the inability to comment on human-authored PRs with suggestions like @copilot do this, and a lack of support for Windows or macOS environments.

Regarding the latter, GitHub Copilot Coding Agent only runs in Ubuntu (Linux) environments within GitHub Actions. It does not currently support Windows or macOS runners. That means when it tries to build or test a .NET MAUI app that's designed to run on Windows or macOS, it can't do platform-specific builds or validations because those OSes aren't available in the agent's runtime. "This is unfortunately a big need for .NET MAUI, as the product targets Windows and iOS," the team noted.

Still, the team sees potential for Copilot to continue expanding its role. "Over the next several months, we should have a better picture of Copilot Coding Agent's full impact on the product."