GitHub Copilot adds an interactive upgrade canvas for .NET modernization

Microsoft has introduced an interactive upgrade canvas for GitHub Copilot, designed to make legacy modernization transparent and steerable for developers. Announced in a July 9, 2026 post on the .NET blog and reported by InfoWorld, the canvas provides a live view of the Copilot upgrade agent's workflow as it assesses, plans, and executes .NET application upgrades.

Modernizing legacy enterprise applications has traditionally been tedious manual work. The Copilot upgrade agent aims to automate the grind. It starts by assessing the target application to answer the questions a human would ask first: which version of .NET is the application targeting, which NuGet packages need updates, which breaking API changes have to be resolved, which projects in a solution can be upgraded independently, and what order the work should happen in.

From that assessment the agent generates a structured, multi-step upgrade plan and breaks the modernization into actionable tasks. It also adapts as it goes, reacting to problems it discovers along the way such as dependency conflicts and build failures.

The more interesting part is the canvas itself. Coding agents that tackle long multi-step jobs have tended to run as black boxes: you hand over a repository, wait, and read a diff at the end. Microsoft describes the canvas as offering a live view of the modernization workflow as it unfolds, in a single place rather than scattered across editors, terminals and pull requests. That means engineers can watch the plan take shape and step in to redirect the agent while the work is happening instead of after it.

Microsoft has made the upgrade agent available across several environments:

• The GitHub Copilot app hosts the interactive upgrade canvas with the full assessment and execution timeline.
• Visual Studio builds it into Solution Explorer — right-click a solution or project and select Modernize.
• Visual Studio Code supports it through the GitHub Copilot upgrade extension from the marketplace.
• GitHub Copilot CLI exposes it via a plugin, so assessment, planning and execution can run from the terminal.

For developers learning to work alongside AI agents, the canvas is worth noting less for what it upgrades than for how it behaves. A .NET version bump is a well-bounded problem with a clear finish line — exactly the kind of task an agent should be good at, and exactly the kind where a silent failure buried in a 200-file diff is expensive to unpick. Putting the agent's plan on screen before and during execution is a bet that the useful unit of trust is the plan, not the patch.

That bet reflects a broader shift in how agentic tools are being designed in 2026: away from blind execution, and toward agents that explain what they intend to do while there is still time to say no.