Installing Continue on VS Code: Build Your Own Premium AI Assistant to Replace GitHub Copilot

Quick Start: Run Continue in Just 5 Minutes

Want to try it out right away? It only takes 3 steps for AI to start helping you code directly in VS Code without spending a dime on licensing fees.

1. Install the Extension: Open VS Code, press Ctrl+Shift+X, search for Continue, and select Install.
2. Choose a Model: Click the Continue icon in the sidebar. If you don’t have an API Key, choose the models available in the Free Trial to experience it first.
3. Test it out: Highlight a block of code and press Ctrl + L (Windows) or Cmd + L (Mac). Then, try asking: “Optimize this code for me.”

And that’s it. However, to truly replace GitHub Copilot in real-world projects, you need to know how to configure it more deeply below.

Why Use Continue Instead of GitHub Copilot?

GitHub Copilot is excellent, but the $10/month cost (about 2.5 million VND/year) is a barrier for many. More importantly, many companies currently ban Copilot due to concerns about source code being uploaded to Microsoft’s servers.

Continue completely solves this issue. It is an open-source framework that allows you to flexibly change the AI “brain.” You can use GPT-4o for complex logic or run entirely offline with Llama 3 via Ollama for absolute company code security.

Detailed Configuration: Connecting the AI Brain

Continue acts as the interface, while the real power lies in the Large Language Models (LLMs) you connect to it.

1. Run Local AI with Ollama (Free and Private)

If your machine has 16GB of RAM or more, prioritize this method. Your code will never leave your personal computer.

• Download Ollama at ollama.com and install it.
• Open your terminal and type: ollama run llama3.
• In Continue, click the gear icon to open the config.json file and add the following snippet:

{
  "title": "Ollama Llama 3",
  "model": "llama3",
  "type": "ollama"
}

2. Using High-Speed APIs (Claude 3.5, Groq)

For low-spec machines, using an API is the optimal choice. I especially recommend Groq because of its extremely fast response time, often exceeding 400 tokens/second, meaning you almost never have to wait.

Sample configuration in config.json:

{
  "models": [
    {
      "title": "Claude 3.5 Sonnet",
      "provider": "anthropic",
      "model": "claude-3-5-sonnet-20240620",
      "apiKey": "YOUR_API_KEY"
    },
    {
      "title": "Groq Llama 3",
      "provider": "groq",
      "model": "llama3-70b-8192",
      "apiKey": "YOUR_GROQ_API_KEY"
    }
  ]
}

Leveraging Context and Slash Commands

What sets Continue apart from standard chat is its ability to understand your entire project through Context.

Using Context Providers with the @ Key

When chatting, type @ to specify the data source the AI should read:

• @Files: Specify a specific file for the AI to analyze for errors.
• @Codebase: Continue will scan the entire project. You can ask: “Where is the payment handling function?” without searching manually.
• @Docs: Paste a documentation link (like React or Tailwind), and the AI will update its knowledge with the latest information to support you.

Quick Commands (Slash Commands)

Save typing time with / commands:

• /edit: Highlight code and type /edit add try-catch, and the AI will apply the edit directly to the file.
• /comment: Automatically generate standardized docstrings and annotations.
• /test: Write Unit Tests automatically. This feature has helped me reduce boilerplate code writing time for testing by 30%.

Real-world Experience After 6 Months of Use

Don’t use a “sledgehammer to crack a nut.” For simple function explanations, Llama 3 running locally is more than enough. Save your GPT-4o or Claude 3.5 API quota for complex system refactoring tasks.

Leveraging shortcuts is key to increasing productivity. Make it a habit to use Ctrl + L to chat and Ctrl + I to edit code in place. Minimizing mouse usage will keep your train of thought uninterrupted.

Finally, set up the systemPrompt in your configuration file. You can request the AI to always reply in English or follow specific project coding conventions for more consistent results.

Comparison: Continue vs GitHub Copilot

Mastering an open-source tool like Continue helps you avoid dependency on any single provider. Although configuring a JSON file might feel unfamiliar at first, in return, you get an AI assistant that is fully personalized to your liking.

Good luck optimizing your development workflow and coding more smoothly with Continue!