---

📦 StackScribe: AI-Powered Code Documentation

StackScribe is a next-generation CLI tool and NPM module that automatically generates professional, high-quality comments for your code. It scans your codebase, identifies functions and APIs, and leverages powerful Large Language Models (LLMs) to write detailed explanations—saving you hours of manual documentation.

---

⚠️ Under Development

Currently supports JavaScript and TypeScript. Support for additional languages like Python, Java, and more is planned in future releases.

---

🚀 Usage

Getting started with StackScribe is quick, whether as a CLI tool or a module in your project.

Install the package

npm install stackscribe

Commands Guide

npx stackscribe --help

# To see Version
npx stackscribe --version

1️⃣ Command-Line (CLI) Usage

Step 1: Configure Your API Key (One-time setup)

Ollama runs locally and requires no key. For other providers, set your API key:

# Groq example
npx stackscribe config --provider groq --apiKey YOUR_GROQ_API_KEY

# OpenAI
npx stackscribe config --provider openai --apiKey YOUR_OPENAI_API_KEY

# Gemini
npx stackscribe config --provider gemini --apiKey YOUR_GEMINI_API_KEY

Step 2: Run on Your Codebase

Generate comments for a single file or an entire directory:

# Run on './src' directory using Groq
npx stackscribe run --path ./src --provider groq --model llama3-8b-8192

# Run on a Python file using Gemini
npx stackscribe run --path ./my_script.py --provider gemini --model gemini-1.5-pro

# Use the default provider from your config
npx stackscribe run --path ./src/myFile.js

---

2️⃣ Programmatic (Module) Usage

Integrate StackScribe directly into scripts, CI/CD pipelines, or other tools.

Step 1: Install the package

npm install stackscribe

Step 2: Import and use

import { main } from "stackscribe";

// Run on './src' directory using default provider
main("./src");

// Specify a provider
main("./src", "gemini");

// Specify provider and model
main("./src", "groq", "llama3-8b-8192");

---

📝 About the Project

StackScribe is designed to be simple, powerful, and easy to integrate into modern development workflows.

Key Features

• 🤖 Intelligent Code Analysis Uses AST parsing to accurately detect all functions and API calls in JavaScript/TypeScript code, ignoring irrelevant sections. Other languages are fully processed.

• 🔗 Multi-Provider LLM Support Choose from OpenAI, Google Gemini, Groq, or run offline with Ollama. Specify models for each provider for maximum control.

• ✍️ High-Quality Comment Generation Generates professional, list-style explanations of function logic, purpose, and expected inputs/outputs.

---

Example Comment Output

/**
 * Handles user login by validating credentials and issuing a JWT.
 * 1. Extracts email and password from the request body.
 * 2. Validates credentials against the database.
 * 3. On success, generates access and refresh tokens.
 * 4. Returns the tokens to the client.
 * @param {object} req - Express request object
 * @param {object} res - Express response object
 */
function loginUser(req, res) {
   // ... function logic
}

---

Project Architecture

stackscribe/
├─ bin/
│  └─ stackscribe.ts       # CLI commands (yargs)
├─ src/
│  ├─ index.ts             # Main logic (CLI + module)
│  ├─ parser.ts            # AST parsing
│  ├─ annotator.ts         # Insert comments
│  ├─ llm/
│  │  ├─ openai.ts         # LLM wrappers
│  │  ├─ gemini.ts
│  │  ├─ groq.ts
│  │  └─ ollama.ts
│  ├─ config.ts            # API keys & config
│  └─ utils.ts             # Helper functions
├─ package.json
└─ README.md

---

Core Technologies

• Code Parsing: @babel/parser, @babel/traverse
• Code Generation: recast (preserves formatting)
• LLM SDKs: openai, @google/generative-ai, groq-sdk, ollama
• CLI Framework: yargs

---

🛠 Development Roadmap

• Phase 1 (MVP): JS/TS support with OpenAI — ✅ Completed
• Phase 2 (Multi-Provider): Gemini, Groq, Ollama integration — ✅ Completed
• Phase 3 (Config): Robust CLI + stackscribe.json for project defaults — 🟡 Partially Completed
• Phase 4 (Multi-Language): Python, Java, etc. support — 🟡 Partially Completed