AI DEV Agents Comparison - by Edukey

An experimental project to compare:

• GitHub Copilot Workspace + VSCode CW plugin + SuperMaven autocomplete,
• Cursor Composer + Cursor Tab autocomplete,
• Aider with Claude 3.5 Sonnet + VSCode & SuperMaven autocomplete,
• Claude Dev (Cline) with Claude 3.5 Sonnet + VSCode & SuperMaven autocomplete,
• "traditional" ChatGPT (o1-mini) + VSCode (copy/paste) & SuperMaven autocomplete

in Front-End development with Next.js, TailwindCSS, MongoDB Node.js Driver, and next-intl internationalization. See what we will build in the end.

Setup

1. Install the following:

   • Cursor Composer (there is a free trial)
   • VSCode with:
     • SuperMaven (also free trial available)
     • Claude Dev (Cline) (open source, your own API key is required)
   • Aider CLI tool (open source, your own API key is required)
   • GitHub Copilot Workspace is in Closed Beta, so you need to sign up for it (I will share a link to my session - you can access it with your GitHub account even if you don't have a GitHub Copilot Workspace access)

2. Clone this repo and use main branch as a starting point for own comparisons.

3. You will find separate comparisons in folowing branches:

   • github-copilot-workspace compares GitHub Copilot Workspace + VSCode CW plugin + SuperMaven autocomplete
   • cursor-composer compares Cursor Composer + Cursor Tab autocomplete
   • aider compares Aider with Claude 3.5 Sonnet + VSCode & SuperMaven autocomplete
   • claude-dev compares Claude Dev (Cline) with Claude 3.5 Sonnet + VSCode & SuperMaven autocomplete
   • chatgpt-o1-copy-paste compares "traditional" ChatGPT + VSCode (copy/paste) & SuperMaven autocomplete & Continue

4. Folder structure:

   • /public/images contains Edukey logo, favicon and hero image in 2 sizes and in JPG and WebP format.
   • .env contains database connection string (you have example.env file to copy and rename).
   • /locales folder with empty en.json and pl.json files.

5. MongoDB Validation Schema: See MongoDB-Validation.md file for the MongoDB validation schema for the waiting-list collection.

Action Plan

We will create a series of videos and blog posts, starting with an introduction to the project and the AI tools we will use, and then we will create a series of comparisons, each comparing one of the AI tools with the others.

1. Introduction and manual copy/paste with ChatGPT (o1-mini or o1-preview or 4o with canvas) and SuperMaven autocomplete.
2. GitHub Copilot Workspace + VSCode CW plugin + SuperMaven autocomplete.
3. ...

Landing Page description

TL;DR: we will build a responsive landing page for a startup, with a hero image, a 2-step form to sign on our "Waiting list", with validation on FE, BE, and in the database, and a toggle language selector.

Detailed description of the landing page we will build, together with desktop and mobile designs (export from Figma), is available in Landing-Description.md file. We will use this descriptions as our base for prompts we send to AI tools.

License

MIT License - I ask only for attribution, but you can use it for any purpose.