BSA-AI - Multi-AI Model CLI SDK

A powerful Node.js CLI-based SDK that allows seamless interaction with multiple AI models including ChatGPT, Google Gemini, Claude, Deepseek, and GitHub Copilot. Built with TypeScript for robust type safety and developer experience.

Features

• Multi-Provider Support: Interact with 5 major AI providers from a single interface

  • OpenAI ChatGPT (GPT-4, GPT-4-turbo, GPT-3.5-turbo)
  • Google Gemini (Gemini 1.5 Pro, Flash)
  • Anthropic Claude (Claude 3.5 Sonnet, Haiku, Opus)
  • Deepseek (Deepseek Chat, Coder)
  • GitHub Copilot

• Persistent Interactive Sessions: Start a session that runs until explicitly closed

• File System Operations: Read, write, edit, and delete files in your project

• Command Execution: Execute shell commands with safety checks

• Consent Prompts: User approval required for file modifications and command execution

• Streaming Responses: Real-time streaming for supported providers

• Session Management: Save and load conversation sessions

• Configuration Management: Easy provider configuration with secure API key storage

Installation

From NPM (after publishing)

npm install -g bsa-ai

From Source

# Clone the repository
git clone <your-repo-url>
cd bsa-ai

# Install dependencies
npm install

# Build the project
npm run build

# Link globally
npm link

Quick Start

1. Configure a Provider

Before using BSA-AI, you need to configure at least one AI provider:

# Configure interactively
bsa-ai config

# Or configure a specific provider
bsa-ai config -p chatgpt

You'll be prompted to enter:

• API key (required)
• Base URL (optional, for custom endpoints)
• Default model (optional)
• Temperature (optional, default: 0.7)
• Max tokens (optional)

2. Start an Interactive Session

# Start with default provider
bsa-ai start

# Start with specific provider
bsa-ai start -p gemini

# Start with custom working directory
bsa-ai start -d /path/to/your/project

3. Interact with the AI

Once in a session, you can:

You: Help me create a Node.js express server
Assistant: I'll help you create a Node.js Express server...

CLI Commands

Session Commands

• bsa-ai start - Start an interactive AI session
  • -p, --provider <provider> - Specify AI provider (chatgpt, gemini, claude, deepseek, copilot)
  • -d, --directory <directory> - Set working directory (default: current directory)

Configuration Commands

• bsa-ai config - Configure a provider interactively

  • -p, --provider <provider> - Configure specific provider

• bsa-ai list - List all configured providers

• bsa-ai default <provider> - Set default AI provider

• bsa-ai info - Show configuration information

• bsa-ai clear - Clear configuration

  • -p, --provider <provider> - Clear specific provider config

In-Session Commands

While in an interactive session, use these commands:

• /exit or /quit - Exit the session
• /clear - Clear conversation history
• /save - Save current session to file
• /history - Display conversation history
• /help - Show help message

Obtaining API Keys

OpenAI (ChatGPT)

1. Visit https://platform.openai.com/api-keys
2. Sign in or create an account
3. Click "Create new secret key"
4. Copy and save your API key

Google Gemini

1. Visit https://makersuite.google.com/app/apikey
2. Sign in with your Google account
3. Click "Create API Key"
4. Copy and save your API key

Anthropic (Claude)

1. Visit https://console.anthropic.com/
2. Sign in or create an account
3. Navigate to API Keys
4. Create a new API key
5. Copy and save your API key

Deepseek

1. Visit https://platform.deepseek.com/
2. Sign in or create an account
3. Navigate to API Keys section
4. Create a new API key
5. Copy and save your API key

GitHub Copilot

1. Requires GitHub Copilot subscription
2. Visit https://github.com/settings/tokens
3. Generate a personal access token with appropriate scopes
4. Use the token as your API key

Project Structure

bsa-ai/
├── src/
│   ├── cli.ts              # CLI entry point
│   ├── index.ts            # Main exports
│   ├── types/
│   │   ├── index.ts        # Type definitions
│   │   └── provider.ts     # Provider interfaces
│   ├── providers/
│   │   ├── chatgpt.ts      # OpenAI provider
│   │   ├── gemini.ts       # Google Gemini provider
│   │   ├── claude.ts       # Anthropic Claude provider
│   │   ├── deepseek.ts     # Deepseek provider
│   │   ├── copilot.ts      # GitHub Copilot provider
│   │   └── index.ts        # Provider factory
│   ├── managers/
│   │   ├── configManager.ts    # Configuration management
│   │   ├── sessionManager.ts   # Session management
│   │   ├── fileManager.ts      # File operations
│   │   ├── consentManager.ts   # User consent handling
│   │   └── commandManager.ts   # Command execution
│   ├── core/
│   │   └── interactiveSession.ts  # Interactive session logic
│   └── utils/
│       └── logger.ts           # Logging utilities
├── dist/                    # Compiled JavaScript
├── package.json
├── tsconfig.json
└── README.md

Programmatic Usage

You can also use BSA-AI as a library in your own Node.js projects:

import {
  createProvider,
  AIProvider,
  SessionManager,
  ConfigManager
} from 'bsa-ai';

// Create a provider
const provider = createProvider(AIProvider.CHATGPT);
provider.configure({
  apiKey: 'your-api-key',
  model: 'gpt-4',
  temperature: 0.7
});

// Use session manager
const configManager = new ConfigManager();
const sessionManager = new SessionManager(configManager);

await sessionManager.initializeSession(AIProvider.CHATGPT);
const response = await sessionManager.sendMessage('Hello!');
console.log(response);

File Operations

The SDK includes a file manager that can:

• Read files: Access file contents in your project
• Write files: Create new files with content
• Edit files: Modify existing file contents
• Delete files: Remove files (with consent)
• List files: Browse directory contents

All file operations require user consent through interactive prompts.

Command Execution

Execute shell commands safely:

• Commands are executed in the specified working directory
• Dangerous commands (rm -rf, format, etc.) require explicit consent
• Output (stdout/stderr) is captured and displayed
• Error handling for failed commands

Consent System

BSA-AI uses a three-option consent system similar to Claude CLI for enhanced control:

When the AI requests to edit, create, or execute files/commands, you'll be prompted with:

• Yes - Approve this single operation only
• Yes for current session - Auto-approve all similar operations for the remainder of the session
• No - Reject this operation

Session-Based Auto-Approval

• File operations and command executions have separate session approval settings
• Choosing "Yes for current session" remembers your preference until:
  • You exit the session
  • You use the /clear command (resets both history and consent preferences)
• Auto-approved operations show: ✓ Auto-approved (session setting)

Example

⚠️  File Operation Requested:
Type: WRITE
Path: server.js
Content preview: const express = require('express')...

? Do you want to proceed?
  Yes
> Yes for current session
  No

After selecting "Yes for current session", all subsequent file operations in that session will be automatically approved.

Security Considerations

• API keys are stored securely using the conf package
• Configuration files are stored in OS-specific config directories
• All file modifications require user consent
• Dangerous commands are flagged and require approval
• No API keys are logged or displayed

Configuration File Location

BSA-AI stores configuration in platform-specific locations:

• macOS: ~/Library/Preferences/bsa-ai-nodejs/
• Windows: %APPDATA%\bsa-ai-nodejs\Config\
• Linux: ~/.config/bsa-ai-nodejs/

To view config location: bsa-ai info

Development

Building

npm run build

Development Mode (watch)

npm run dev

Testing

npm test

Troubleshooting

Provider not configured error

Run bsa-ai config -p <provider> to configure the provider with your API key.

API key errors

Verify your API key is correct and has not expired. Reconfigure using bsa-ai config.

Connection errors

Check your internet connection and verify the provider's API is accessible.

Permission errors

Ensure the CLI has permission to read/write files in your working directory.

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests for new functionality
5. Submit a pull request

License

MIT License - see LICENSE file for details

Acknowledgments

• Built with TypeScript
• Uses official SDKs from OpenAI, Google, and Anthropic
• Powered by Commander.js, Inquirer.js, and Chalk

Support

For issues and feature requests, please open an issue on GitHub.

Note: This SDK requires Node.js 16.0.0 or higher.