contaigents/CHAT_COMMAND.md

Modular AI Content Ecosystem with Audio Generation

# Chat Command

The Contaigents CLI now includes a `chat` command that provides a conversational AI interface for interactive assistance.

## Usage

```bash
contaigents chat [options]
```

## Options

- `-p, --provider <provider>` - AI provider to use (openai, anthropic, google, etc.)
- `-m, --model <model>` - Model to use for generation
- `-t, --temperature <temperature>` - Temperature for response generation (0.0-1.0)
- `--max-tokens <maxTokens>` - Maximum tokens for response
- `--system-prompt <systemPrompt>` - Custom system prompt for the chat session
- `--non-interactive` - Run in non-interactive mode (for testing)

## Examples

### Start an interactive chat session:
```bash
contaigents chat --provider openai
```

### Use a specific model and temperature:
```bash
contaigents chat --provider openai --model gpt-4 --temperature 0.8
```

### Use a custom system prompt:
```bash
contaigents chat --provider openai --system-prompt "You are a coding assistant specialized in JavaScript"
```

### Non-interactive mode (for testing):
```bash
contaigents chat --provider openai --non-interactive
```

## Interactive Commands

Once in the chat session:
- Type your messages and press Enter to send
- Type `exit` or `quit` to end the session
- Press `Ctrl+C` to force quit

## HTTP API Endpoints

The chat functionality is also available via HTTP API:

### Quick Chat (Single Message)
```bash
POST /api/chat/quick
Content-Type: application/json

{
  "message": "Hello, what can you help me with?",
  "options": {
    "provider": "openai",
    "temperature": 0.7
  }
}
```

### Session-Based Chat

#### Create Session
```bash
POST /api/chat/session
Content-Type: application/json

{
  "systemPrompt": "You are a helpful assistant",
  "provider": "openai"
}
```

#### Send Message to Session
```bash
POST /api/chat/message
Content-Type: application/json

{
  "sessionId": "chat_123456789_abcdef",
  "message": "What is 2+2?",
  "options": {
    "provider": "openai"
  }
}
```

#### Get Session Messages
```bash
GET /api/chat/session/{sessionId}/messages
```

#### List All Sessions
```bash
GET /api/chat/sessions
```

#### Delete Session
```bash
DELETE /api/chat/session/{sessionId}
```

## Supported Providers

The chat command works with any configured LLM provider:
- **OpenAI** - GPT models (gpt-4, gpt-3.5-turbo, etc.)
- **Anthropic** - Claude models
- **Google** - Gemini models
- **And more** - Any provider configured in your system

## Configuration

Make sure you have at least one LLM provider configured before using the chat command. You can configure providers through the web interface or by manually editing the configuration files in `.config/llm_config/`.

## Features

- **Conversation Memory** - Maintains context across messages in a session
- **Multiple Providers** - Works with any configured LLM provider
- **Flexible Configuration** - Customizable temperature, tokens, and system prompts
- **Session Management** - Create, manage, and delete chat sessions
- **Error Handling** - Graceful error handling and user feedback
- **Interactive Interface** - User-friendly command-line chat experience

## Future Enhancements

This basic conversational agent is designed to scale into a context-aware agent that can:
- Create and modify markdown, MDX, and text files
- Understand project context and structure
- Perform file operations and content management
- Integrate with the broader Contaigents ecosystem