Exploring AI Code Editors: A Hands‑On Review of Copilot, Cursor, CodeGeeX and More

Overview of AI Code Editors

The piece introduces a set of AI‑assisted programming tools that integrate with local editors either as plugins or as standalone applications, collectively referred to as “AI code editors.”

Main Tools Covered

GitHub Copilot

Copilot, launched as a technical preview in 2021 and commercialized in 2022, is co‑developed by GitHub and OpenAI. It works as a plugin for VS Code, JetBrains IDEs and other mainstream editors, offering code completion, intelligent dialogue, and project‑level editing. Pricing is $10 USD per month, with free access for students and a one‑month trial for new users.

Cursor

Cursor is a newer, standalone editor built on the VS Code engine, gaining popularity in the second half of 2024. In addition to basic completion, it provides batch completion, next‑step prediction, multi‑file modifications and an advanced chat interface. Its UI and feature set closely resemble Copilot’s, allowing skills learned in one tool to transfer to the other. Subscription costs $20 USD per month with a two‑week free trial.

CodeGeeX

Developed by Zhipu AI, CodeGeeX supports multiple programming languages via plugins for VS Code and JetBrains IDEs. It is free for individual users, while enterprises can opt for a private‑deployment solution.

Other Notable Editors

Additional offerings include Alibaba’s Tongyi Lingma, ByteDance’s MarsCode, Tabnine, Windsurf and various plugins from major Chinese internet firms, all of which are free for personal use and share similar feature sets.

Using an AI Code Editor – Example with GitHub Copilot

Preparation

Install an editor (recommended: VS Code).

Create a GitHub account.

Install required extensions:

VS Code Chinese language pack

GitHub Copilot extension

Python extension

Configure the editor and extensions.

Set up a Python runtime environment.

Further details are available in the referenced book “AI‑Assisted Programming Intro.”

Core Feature 1: Code Completion

Code completion is the earliest and most basic capability of tools like Copilot. As you type, the model suggests inline code snippets, ranging from single statements to whole function implementations. The suggestions appear in gray italic text and are not yet part of the file.

Press Tab to accept a suggestion and insert it as real code.

Press Esc to dismiss the suggestion.

Continue typing manually if the suggestion is unsuitable.

The model keeps generating new suggestions based on subsequent input.

The low interaction cost makes this feature highly efficient for repetitive coding tasks.

💡 A common tip is to write a comment describing the desired behavior and let Copilot generate the corresponding code.

Limitations: a single completion can only produce a limited amount of code, making it unsuitable for generating entire programs or editing multiple files simultaneously.

Core Feature 2: Inline Chat

Activating the chat dialog at the cursor position lets you ask questions or request code changes. Generated code appears in a light‑green background; accepting it inserts the code into the file.

This mode is useful for on‑the‑fly modifications that the basic completion cannot handle.

Core Feature 3: Sidebar Chat

The sidebar acts like an integrated ChatGPT, answering project‑wide questions, generating or refactoring code across multiple files, and leveraging large models such as GPT‑4o.

New Feature: Project Editing

Copilot’s recent project‑editing capability can add new files, implement new features, or modify existing ones across an entire codebase, offering a more streamlined workflow than the sidebar for multi‑file operations.

Additional Helper Functions

Run the current Python file and view output in the terminal.

Run selected Python code with Shift+Enter.

Generate terminal commands from natural‑language prompts.

Explain selected code or terminal output.

Perform code review with automated quality suggestions.

Conclusion

The guide demonstrates that AI code editors, especially GitHub Copilot, can significantly accelerate development through low‑friction code completion, interactive chat, and project‑wide editing. While experienced programmers can quickly adopt these tools, beginners may need additional guidance, which the author promises to provide in the next installment.