CLIP - Command Line Interface for Prompting

Note

This project is currently a work in progress. Feel free to star the repository to stay updated on new features and improvements. Feedback and suggestions are always welcome!

CLIP is a command-line interface to chat with LLM models that supports OpenAI's Response API.

Features

• Conversation Management: Tracks conversation context for follow-up questions
• Context Pause/Resume: Pause conversations and resume them later with the /continue command
• Conversation-Based History: Responses are grouped by conversation for better organization
• Web Search Capability: Optionally enable web search for up-to-date information
• Command Interface: Intuitive slash commands with autocomplete
• Response ID Tracking: Uses OpenAI's response IDs for conversation continuity

Technical Overview

CLIP is built as a personal tool using:

• Node.js
• OpenAI's Responses API
• Terminal interface with readline

Commands

The application supports these commands:

• /help - Show help information
• /continue - Resume conversation with the most recent response
• /continue n - Continue from response #n in the current conversation (e.g., /continue 2)
• /history - View the response history for the current conversation
• /history --all - View response history across all conversations
• /new - Start a new conversation with fresh context
• /debug - Show current conversation state (for troubleshooting)
• /clear - Clear the screen
• /exit - Exit the application

Conversation Modes

CLIP offers two conversation modes:

1. Standard mode - Regular conversation with GPT
2. Web search enabled - Uses OpenAI's search tool for current information

OpenAI Responses API Integration

CLIP uses OpenAI's Responses API for conversation management, providing:

• Conversation continuity through response IDs
• Streaming response handling
• Support for tool usage (such as web search)
• Multi-turn conversation handling

Debugging

CLIP includes a debugging system controlled through environment variables:

# Enable basic info level debugging
DEBUG=clipai node index.js

# Enable verbose level (more detailed output)
DEBUG=clipai:verbose node index.js

# Enable debug level (maximum detail)
DEBUG=clipai:debug node index.js

Each debugging level provides progressively more information about API interactions and application state.

Data Storage

CLIP stores conversation data in a dedicated directory:

~/.clipai/
├── conversations/        # Conversation data organized by ID
├── responses/            # Response history organized by conversation
├── state.json            # Current application state
└── config.json           # User configuration

License

MIT