I was wondering if there's a decent AI agent that helped with coding but can be run locally. After a quick test, I managed to get it to work with the following stack:

• Ollama
• DeepSeek Coder
• Continue VS Code Extension
• VS Code

To set it up, I follow the steps below:

1. First, I install Ollama on my machine using winget: winget install Ollama.Ollama.
2. Then, download DeepSeek Coder to run in Ollama. https://ollama.com/library/. In my case, I just use the smallest model which is 1.3b paramaters: ollama run deepseek-coder:1.3b.
3. Next, I search for and install "Continue" VS Code extension by continue.dev.
4. To configure the extension, make sure Ollama is running on http://localhost:11434. If it's not, run ollama serve.
5. In my case, the "Continue" extension was unable to detect Ollama. I had to setup the configuration to autodetect. To do that, I open the Configs menu of the extension. Click the setting/gear icon next to Local Config and paste the following config:

name: Local Config
version: 1.0.0
schema: v1
models:
  - name: Autodetect
    provider: ollama
    model: AUTODETECT
    roles:
      - chat
      - edit
      - apply
      - rerank
      - autocomplete

Then, the LLM works locally, but it will use a lot of CPU and memory.