MyPerplexity CLI

An AI-powered interactive command-line interface that answers questions by combining real-time web search with advanced AI reasoning. Built with Genkit and Tavily APIs.

Learning Project: Developed as part of Big School's Master de desarrollo con IA program.

Features

• 🔍 Real-time Web Search: Uses Tavily API to search current information from the web
• 🤖 AI-Powered Responses: Leverages Google's Gemini AI to synthesize search results into comprehensive answers
• 💬 Interactive CLI: User-friendly command-line interface with spinners and colored output
• 📚 Source Citations: Automatically includes URLs and sources in responses
• 💾 Chat History: Maintains conversation history during the session

Prerequisites

• Node.js v18+
• API Keys:
  • Google API Key for Gemini AI
  • Tavily API Key for web search

Installation

1. Clone the repository

   git clone <repository-url>
   cd 09-master-ia-cli-genkit

2. Install dependencies

   npm install

3. Set up environment variables

   cp .env.example .env

   Then edit .env and add your API keys:

   TAVILY_API_KEY=your_tavily_key_here
   GOOGLE_API_KEY=your_google_api_key_here
   

Usage

Interactive Mode (CLI)

npm start

This launches the interactive CLI where you can ask questions:

💬 Ask a question (or type "exit" to quit): What is the capital of France?

🧁 Thinking...
✅ Answer Received!

Answer:
The capital of France is Paris. Paris is the largest city in France and serves as the 
political, cultural, and economic center of the country...

Sources:
[1] https://example.com/france-capital
[2] https://example.com/paris-info

Development Mode (Genkit UI)

npm run dev

This also runs the interactive CLI and starts the Genkit Developer UI on http://localhost:4000 for development and debugging.

To access the UI:

1. Click on the http://localhost:4000 link in the terminal, or
2. Open your browser and navigate to http://localhost:4000

The Genkit UI allows you to:

• Test flows and prompts interactively
• View execution traces and AI model interactions
• Debug tool calls and API responses
• Monitor performance metrics

Screenshot

Interactive CLI in action - Ask questions and get AI-powered answers with web sources

Project Structure

09-master-ia-cli-genkit/
├── index.js              # Main entry point - CLI setup and interaction loop
├── src/
│   ├── agent.js         # Chat agent with search tool configuration
│   └── search.js        # Tavily web search wrapper
├── .env.example         # Environment variables template
└── README.md            # This file

How It Works

1. User Input: User enters a question in the CLI
2. Search: The chat agent uses Tavily's searchWeb tool to find current information
3. AI Processing: Google Gemini AI synthesizes search results with its knowledge
4. Response: A comprehensive answer is formatted with citations and sources
5. History: The exchange is stored in the chat session history

Architecture

• index.js: Manages the CLI interface using Node.js readline module
• agent.js: Defines the Genkit agent with the search tool and AI prompt
• search.js: Wraps the Tavily API with proper configuration options

Tech Stack

Technology	Purpose
Genkit	AI framework for building agentic apps
Google Gemini AI	Language model for reasoning
Tavily API	Real-time web search
Zod	TypeScript-first schema validation
Chalk	Terminal string styling
Ora	Elegant terminal spinners
dotenv	Environment variable management
Readline	CLI input/output (Node.js built-in)

References

Framework & AI APIs

• Genkit Documentation
  • Chat & Streaming
  • Tool Calling
• Google AI Studio
• Tavily API

Dependencies

• @tavily/core - Tavily web search client
• Chalk - Terminal styling
• Ora - Progress spinners
• dotenv - Environment variables
• Zod - Schema validation

Node.js APIs

• Readline Module - Interactive CLI

Learning Source

This project was developed as part of the Master de Desarrollo con IA by Big School, a comprehensive program covering AI development with modern tools and frameworks.

Completion Date: February 9, 2026

Possible Future Enhancements

• Use native Node.js fetch instead of Tavily package dependency
• Add conversation persistence (save/load chat history)
• Implement multi-turn context awareness
• Add support for different AI models
• Web dashboard interface alongside CLI