Anyany.js – Generative AI CLI for Debugging & Testing

A Node.js-based CLI tool that leverages generative AI to assist with debugging, testing, and other quality-related tasks
Supports both local models (via Ollama) and cloud-based models (such as OpenAI and OpenRouter)

✨ Fitur Utama

🎯 Multi-Model & Multi-Provider

Seamlessly connect with various AI providers:

• OpenAI
  Models like GPT-4o, GPT-4 Turbo, etc

• OpenRouter
  Access to models like Google Gemini, Meta Llama, Mistral, and more

• Ollama
  Run local models such as Llama 3, Gemma 2, etc

---

🤖 Advanced Multi-Agent System

Leverage collaborative AI power with multiple modes:

• Debate Mode
  Two or more agents debate to provide diverse perspectives on a topic

• Pipeline Mode
  The output of one agent becomes the input for the next, refining the response step-by-step

• Consensus Mode
  Multiple agents respond independently, and their answers are synthesized into a comprehensive result

---

💾 Persistent Session Management

Never lose track of your work:

• Sessions are saved automatically.
• Reload previous sessions to resume conversations or review history
• View, manage, and delete sessions via an interactive menu

⚡ Powerful Batch Processing

Automate repetitive tasks efficiently:

• File Processing
  Run the same task on multiple queries from .txt or .json files

• Model Comparison
  Test the same query across different AI models to compare performance and results

---

⚙️ Flexible Custom Tasks

Use built-in prompt templates or create your own:

• Leverage pre-defined prompts like bug_analyst, test_data_generator, and scenario_priority
• Add custom instructions at runtime to tailor tasks to your specific needs

---

📊 Versatile Export Options

Save your results in the format you need:

• Export to Plain Text (.txt), Markdown (.md), or JSON (.json)
• Generate clean and stylish HTML reports (.html) with enhanced Markdown conversion

---

🏠 Dynamic Ollama Integration

Automatically detects Ollama models installed on your local system

Installation

1. Clone the repository:

   git clone git@github.com:modalqa/anyany.js.git
   cd anyany.js

2. Install dependencies:

   npm install

3. (Optional) Install showdown for better HTML export:

   npm install showdown

4. Copy .env.example to .env and fill in the required variables:

   cp .env.example .env

   Then, open the .env file and insert your API key

Configuration

Open the .env file and set your API keys. You only need to fill in the keys for the services you intend to use

# Required for OpenAI models
OPENAI_API_KEY="sk-..."

# Required to access models like Gemini, Llama, etc. via OpenRouter
OPENROUTER_API_KEY="sk-or-..."

Usage

Run the application from your terminal:

node agent.js

You will be greeted with an interactive main menu. Simply choose the option you need:

• 🔥 Quick Query: Ask a single quick question to the AI model of your choice
• 🤖 Multi-Agent Mode: Use collaborative modes like Debate, Pipeline, or Consensus
• ⚙️ Custom Task: Use predefined prompt templates (e.g., bug analysis)
• 💾 Session Management: View, load, create, or delete your conversation sessions
• ⚡ Batch Processing: Process multiple queries from a file or compare model outputs
• ❌ Exit: Exit the application.

The results can be displayed in the terminal or exported to files in the output/ folder

---

Example Use Case

Bug Analysis with Pipeline Mode

Use Multi-Agent Mode → Pipeline:

• Agent 1 (GPT-4o): Analyzes error logs to identify potential root causes
• Agent 2 (Llama 3 70B): Builds upon Agent 1’s analysis to write detailed reproduction steps
• Agent 3 (Gemini 1.5 Flash): Summarizes the business impact and provides suggested fixes based on the reproduction steps

Generate Test Data

Use Custom Task → test_data_generator Enter a description of the data you need, and the AI will generate JSON-formatted test data including valid cases, invalid cases, and edge cases

---

Model Comparison

Use Batch Processing → Repeat Query with Different Models Enter a single question, such as:

"Explain the key differences between Playwright and Cypress"

The script will send the query to all configured providers and present a comparison report of the responses

Demo

Note: The above images may not reflect all the latest features

Contribution

Pull requests and issues are very welcome!
Feel free to contribute and help improve the project

---

Copyright (c) modalqa