Cursor Review: What Surprised Me After 30 Days of Use

LakshmanLakshman
3 min read

I’ve been using Cursor—the AI-powered code editor—for over a month now, and it has quickly become my go-to tool for intelligent coding assistance. After evaluating it alongside GitHub Copilot Business, I’ve found Cursor to be more versatile and context-aware, especially for mid-sized tasks.

Here’s a breakdown of what stood out to me 👇


What I Like About Cursor

💻 1. Familiar UI for VS Code Users

Cursor is built on top of VS Code, so there’s no learning curve if you’re already familiar with the environment. It supports all the usual extensions and themes too.

Switching from VS Code to Cursor felt seamless.


✨ 2. Great for Refactoring, Fixes, and Prototypes

Whether you're:

  • Refactoring legacy code

  • Debugging small issues

  • Rapidly building a POC

Cursor handles it well with focused and context-rich suggestions.


🧠 3. Powerful Context Awareness

This is where Cursor truly outshines Copilot. You can include:

  • 🔹 Files, folders

  • 🔹 Open tabs

  • 🔹 Past prompts

  • 🔹 Recent changes

  • 🔹 Code snippets, URLs

  • 🔹 Documentation

  • 🔹 Terminal state

  • 🔹 .cursorrules file

This leads to incredibly relevant suggestions and edits.


🛠️ 4. Background Agents

Cursor runs intelligent agents in the background that can:

  • Perform small refactors

  • Execute medium sized task

  • Even create PRs based on suggestions

It uses “max mode” for deeper reasoning (it uses a larger context window and advanced models).


💬 5. Natural Language to Terminal

Cursor’s terminal lets you type plain English and generates shell commands. For example:

“Install React and Tailwind with pnpm” ➜ generates and runs the exact commands.


⚙️ 6. .cursorrules for Prompt Automation

Think of .cursorrules as reusable prompt macros or automation definitions. You can define:

  • Refactor rules

  • Test case templates

  • Coding standards

  • Repetitive review queries


🔌 7. Extended Model Support

Cursor lets you switch to:

  • Grok-4

  • Deepseek models

  • GPT-4 (Max Mode)

  • Other high-context models as needed

This gives you more control over model behavior and performance.


🤝 8. Slack, GitHub, Figma & MCP Integration

You can ask code questions directly from Slack or integrate feedback workflows with:

  • GitHub PRs

  • Figma designs (via MCP)

  • Terminal or CI/CD processes, etc.


🐛 9. Bugbot for Code Reviews

Cursor’s built-in Bugbot scans your code and:

  • Suggests PR feedback

  • Identifies common security issues

  • Offers inline comments during review


⚠️ What Could Be Better

While Cursor is powerful, it’s not perfect. A few rough edges include:

❌ 1. Fails at times

At times, large-scale edits or multi-file changes sometimes error out or hang. It's not very robust.

❌ 2. Hallucinations Still Happen

Cursor is relatively accurate, but hallucinations (confident about wrong answers) still occur. Especially:

  • Syntax errors

  • Incorrect assumptions in logic, etc


💭 Final Thoughts

If you’re a developer who values:

  • Deep code understanding

  • Workflow automation

  • Pro-level refactoring

  • Model flexibility

Then Cursor is well worth trying. It feels more like a pair programmer than just a code completer.



0
Subscribe to my newsletter

Read articles from Lakshman directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Lakshman
Lakshman