Nota AI

Nota AI is a command-line tool for developers that automatically generates code comments using locally hosted Ollama models.
All processing happens on your machine: no APIs, no cloud data transfer.

Features

• Generates Google-style DocStrings and (optionally) inline comments for:
  • complex logic,
  • algorithms,
  • state changes,
  • error handling,
  • configurations.
• Automatic language detection:
  Python, JavaScript, TypeScript, C++, Java, PHP, Go, C#
• Handles large files (1000–5000+ lines) by splitting them into manageable chunks.
• Configurable via .notarc.json (model, commenting style, etc.).

Installation

Install Nota AI globally using npm:

npm install -g nota-ai

Prerequisites

1. Install Ollama.
2. Pull and run a compatible model, for example:

   ollama pull llama3
   ollama run llama3

3. Ensure Ollama is running locally (default: http://localhost:11434).

Usage

Generate comments for a code file:

nota comment <input_file> [--config <config_file>]

• <input_file> — path to the source code file (e.g., src/index.py)
• --config — optional path to config file (default: .notarc.json)

Example:

nota comment src/main.py

Show help:

nota --help

Configuration

Create a .notarc.json file in the root of your project to customize behavior:

{
  "nota": {
    "ai": {
      "model": "llama3",
      "temperature": 0.5
    },
    "commenting": {
      "includeLineComments": false,
      "includeDocstrings": true,
      "overwriteExisting": false
    }
  }
}

Options

• nota.ai.model — Ollama model name (e.g., llama3, codellama).
• nota.ai.temperature — creativity/randomness (0.0–1.0, default: 0.5).
• nota.commenting.includeLineComments — enable inline comments (default: false).
• nota.commenting.includeDocstrings — generate DocStrings (default: true).
• nota.commenting.overwriteExisting — overwrite existing comments (default: false).

Handling Large Files

• up to 800 lines → single pass;
• 800–4000 lines → chunks of ~1000 lines;
• 4000+ lines → chunks of ~500 lines.

If the model has a small context window (e.g., 8k tokens), processing large files may be slow or fail.
It is recommended to use models with a larger context window (128k+).

Supported Languages

• Python (.py)
• JavaScript (.js)
• TypeScript (.ts)
• C++ (.cpp)
• Java (.java)
• PHP (.php)
• Go (.go)
• C# (.cs)

Commenting Guidelines

Nota AI generates comments that:

• Use Google-style DocStrings for functions and classes.
• Are written in English.
• Focus on business logic, algorithms, non-trivial calculations, and error handling.
• Avoid describing obvious code (e.g., getters/setters).
• Are concise: 1–2 sentences, professional tone.

Notes

• Ensure the selected Ollama model is running before using Nota AI.
• Performance depends on your system and model’s context window.
• Mixed-language files may be partially supported.
• Existing comments are preserved unless overwriteExisting = true.

Example

nota comment src/main.py

Disclaimer

This project does not include or distribute any AI models.
Nota AI only provides an interface for running locally hosted models via Ollama.

All models (e.g., LLaMA, Mistral, Phi, CodeLlama, etc.) are subject to their original licenses.
Users are responsible for downloading models separately and complying with their respective license terms.

License

MIT