GitHub Copilot

What is it

GitHub Copilot is an AI-powered code completion tool developed by GitHub in partnership with OpenAI. It uses a large language model trained on billions of lines of public code to suggest entire functions, boilerplate code, and algorithms based on your context. It integrates with VS Code, JetBrains IDEs, Neovim, and Visual Studio as a plugin.

My Opinion

GitHub Copilot changed how I write code. Not because it writes code for me, but because it eliminates the “I know the pattern, just let me write it out” friction. I can focus on architecture and logic, letting Copilot handle the implementation details. It’s the baseline that every other AI coding tool is measured against.

The Flow State Multiplier

The real value of Copilot is velocity. When you’re in the zone and need to implement a standard pattern—say, a REST endpoint with input validation—Copilot suggests the entire function. You type the function signature, Copilot fills in the body. You review, accept, and move on.

The time saved on boilerplate is massive. It’s not about replacing thinking; it’s about eliminating typing.

The Context Awareness

Copilot is surprisingly good at understanding context. It reads your open files, looks at your imports, and even understands your project structure. It suggests code that fits your style, not just generic implementations. The more you use it, the better it gets at understanding your patterns.

But this context is limited compared to agentic tools like Claude Code or Cursor. Copilot sees your current file and open tabs; it doesn’t understand your entire codebase or git history.

The “Copy-Paste” Problem

The biggest criticism is that Copilot might reproduce code from its training data without attribution. GitHub has added attribution features, but the legal and ethical concerns remain. For open-source projects, this is a genuine risk. For internal codebases, it’s less of an issue.

The practical reality: most suggested code is generic patterns, not copied verbatim. But the debate continues.

The Overreliance Trap

The danger is becoming dependent. When Copilot suggests code, you accept it. Over time, you might forget how to write certain patterns from scratch. This is a real concern for junior developers—they might never learn the fundamentals because the AI always provides the answer.

Use Copilot as an accelerator, not a crutch. You should be able to write the code yourself; Copilot just makes it faster.

The “Plugin” Limitation

Unlike Cursor (a full IDE fork) or Claude Code (an agentic system), Copilot is just a plugin. It can’t refactor across multiple files, run tests to verify changes, or execute commands. It’s autocomplete on steroids, not an AI pair programmer.

For simple code completion, this is fine. For complex refactoring or multi-file changes, you need more capable tools.

Conclusion

GitHub Copilot is the market standard for AI code completion. It works, it’s reliable, and it integrates everywhere. The productivity gains are undeniable, but use it as an accelerator, not a crutch. For more advanced agentic capabilities—executing commands, project-wide refactoring, context-aware debugging—look at Claude Code or Cursor.