KoanAI – Code suggestions for zen programmers
AI is incredible—but using it as a crutch isn’t. The biggest issues I see with AI code completion tools are:
- They disrupt my flow. When working on complex problems, AI suggestions can be more of a distraction than a help, breaking my train of thought.
- They generate misleading code. The output often looks correct at first glance but can be verbose, inefficient, insecure, or just plain wrong.
- They create overreliance. One of the most critical skills for an engineer is the ability to figure things out independently—AI shouldn’t replace that.
Features (More coming very soon!)
- Koan AI Mentor return concrete and actionable hints that are catagorically based,
- Returns example code snippets that are relavant to your issues.
- Covers performance, readability, structure, and security.
Usage
- Open a code file in VSCode.
- Run "KoanAI: Analyze Code" from the Command Palette (
Cmd + Shift + P or Ctrl + Shift + P) to get hints
KoanAI supports both OpenAI (cloud) and Ollama (local models).
Go to VSCode Settings (Cmd + , or Ctrl + ,).
Search for "KoanAI: LLM Provider" and choose:
openai for OpenAI models (GPT-4, GPT-3.5).ollama for local inference (Llama3, Mistral).
Set the model name:
- OpenAI:
gpt-4-turbo, gpt-3.5-turbo
- Ollama:
llama3, mistral, codellama, gemma, phi
If using OpenAI, set your API key in settings under KoanAI: OpenAI API Key.
- You can get an API key from OpenAI.
Need help or have feature requests? Reach out!
| |
