AI Coding

Background

With the development of AI programming (or AI coding), more and more products have become available, such as Cursor and others. To better understand the various tools, extensions, native editors, cloud editors, and terminals in this space, I have gradually compiled this content.

Currently, most AI Coding or Vibe Coding resources in the community serve as navigation sites. I initially considered creating a navigation site myself, but later felt that navigation sites alone don't provide much value. What sets this resource apart is that I want to include my own "opinionated" content.

Current Landscape

Let's discuss the current landscape. Today, the combination of LLM + MCP + Agent is mainstream. We no longer focus on technical details as deeply as before, shifting gradually from depth to breadth. When developing with open-source frameworks or libraries, we don't need to obsess over specific implementation principles. I used to agonize over whether to use Next.js or Astro, but that's no longer important. We can use GitHub MCP and Git MCP to read open-source code. With the emergence of DeepWiki and Context7 MCP, as programmers, we no longer need to stress about which technology to use or even how they work under the hood.

Native Editors

Native editors refer to applications like VSCode that can be installed on a user's computer.

Cursor

Cursor is a product launched around the same time as ChatGPT, and it shines even brighter with the release of Anthropic's Claude 3.5 Sonnet! I have paid for it, and I believe it is currently the most powerful AI editor in the universe.

Windsurf

Windsurf is a product developed by Codeium, released on November 13, 2024. I saw it on Zhihu and tried it immediately, which was very impressive. At the same time, it reinforced my interest in AI coding.

Trae

Trae supports both Mac and Windows systems and is developed by ByteDance.

Zed

Zed is an editor designed for the future, though Windows support is not yet available.

Other Editors

There are several other noteworthy AI editors:

Extensions

Extensions can be integrated into existing editors (like VSCode) to enhance their AI capabilities.

GitHub Copilot

GitHub Copilot is a code completion tool developed by GitHub (Microsoft) and is built into Visual Studio Code.

Cline Roo

Cline Roo integrates DeepSeek R1 with OpenRouter in Visual Studio Code.

Continue

Continue works in conjunction with Roo Cline and DeepSeek.

Other Extensions

Browser-Based Tools

Browser-based tools don't require installation and can be used directly in a web browser.

v0

v0 is an always-on pair-programmer developed by Vercel.

  • Company: Vercel
  • Type: Browser-based
  • Output Type: Web App
  • Website: https://v0.dev/

Bolt

Bolt can be accessed via bolt.diy or bolt.new and is developed by StackBlitz.

  • Company: StackBlitz
  • Type: Browser-based
  • Output Type: Web App
  • Website: https://bolt.new/

Lovable

Lovable is a superhuman full-stack engineer tool.

Devv Builder

Devv Builder can be thought of as the backend version of Bolt.new / Lovable.

Other Browser-Based Tools

Terminal Tools

Terminal tools enhance the command-line interface by adding AI capabilities.

Claude Code

Claude Code is an agentic coding tool made by Anthropic.

Aider

Aider provides AI capabilities in the terminal.

Warp

Warp is an AI-enhanced terminal.

Other Terminal Tools

AI Software Engineers

MetaGPT

MetaGPT (MGX) is a multi-agent framework and standardized SOP AI software engineer.

Devin

Devin is touted as the first AI software engineer.

Other Types of Tools

Mobile App Development

Frontend Component Generation

Git Tools

  • Generate Commit Messages with AI: Use Copilot in Visual Studio Code to generate commit messages

Infrastructure

Conclusion

AI programming tools are rapidly evolving, with new products and features constantly emerging. These tools cover various aspects from code editing, auto-completion, and application generation to documentation creation, greatly improving development efficiency. With the advancement of LLM and Agent technologies, we can expect more innovative AI programming tools to appear in the future.