A terminal application for interacting with various language models using Langchain.js.
- Support for multiple LLM providers: OpenAI, Anthropic, Google, Ollama, and more
- Interactive terminal mode for chat-like interactions
- Command-line mode for one-off queries
- Ability to load system prompts and context from files or strings
# Clone the repository
git clone https://github.com/varlabz/qterm.git
cd qterm
# Install dependencies
npm install
# Build the application
npm run builddocker pull varlabz/qtermdocker build -t qterm .# Start interactive mode
npm start
# Use CLI mode with input
npm start -- -i "Your query here"
# Read input from stdin
echo "Your query here" | npm start -- -i -
# Select a specific provider and model
npm start -- -p google -m gemini-2.0-flash
# Use a custom base URL for the provider
npm start -- -p openai -b https://your-custom-endpoint.com/v1
# Use an MCP configuration file
npm start -- -c ./mcp-config.json
# List available prompts
npm start -- -L
# Show a specific prompt
npm start -- -P "prompt-name"-p, --provider <value>: Provider (e.g., google, openai, openrouter)-m, --model <value>: Model to use-b, --base-url <value>: Base URL for the LLM provider-i, --input <value>: Input text to process (string, file path, or "-" to read from stdin)-s, --system-prompt <value>: System prompt (string or file path)-c, --mcp-config <value>: MCP config file path-L, --list-prompts: List all available prompts from fabric and awesome directories-P, --prompt <value>: Show a specific prompt by name
qterm includes several built-in tools that extend the capabilities of language models:
Extracts transcripts from YouTube videos, allowing the language model to analyze or summarize video content.
# Example usage in a prompt
"Please summarize this YouTube video: https://www.youtube.com/watch?v=dQw4w9WgXcQ"Gets the current time with optional timezone and formatting options.
# Example usage in a prompt
"What time is it now? And what time is it in Tokyo?"Parameters:
format: Locale format (e.g., 'en-US', 'ja-JP')timezone: Timezone (e.g., 'America/New_York', 'Asia/Tokyo')
Uses Playwright to interact with web pages and retrieve their HTML content, enabling the language model to analyze web content.
# Example usage in a prompt
"What is the current headline on example.com?"Commands:
fetch: Get web page content as HTMLlist: List open browser tabs
Note: The Playwright tool requires a Chrome instance running with remote debugging enabled on port 9222.
# Interactive mode
docker run -it --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm
# CLI mode with input
docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -i "Your query here"
# Read input from stdin
echo "Your query here" | docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -i -
# With specific provider and model
docker run -it --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -p google -m gemini-2.0-flash
# Use a custom base URL for the provider
docker run -it --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -p openai -b https://your-custom-endpoint.com/v1
# Use an MCP configuration file (mounted from host)
docker run -it --rm -v $(pwd)/.key:/app/.key:ro -v $(pwd)/mcp-config.json:/app/mcp-config.json:ro varlabz/qterm -c /app/mcp-config.json
# List available prompts
docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -L
# Show a specific prompt
docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -P "prompt-name"You can create shell aliases to make qterm easier to use in your daily workflow. Aliases allow you to run qterm with your preferred settings using shorter commands.
You can create aliases for both npm and Docker approaches.
Add these lines to your ~/.bashrc, ~/.bash_profile (Bash), or ~/.zshrc (Zsh):
# Basic qterm alias
alias qt='npm start --prefix /path/to/qterm'
# qterm with specific provider and model
alias qtg='npm start --prefix /path/to/qterm -- -p google -m gemini-2.0-flash'
alias qto='npm start --prefix /path/to/qterm -- -p openai -m gpt-4o'
# qterm with custom base URL
alias qtc='npm start --prefix /path/to/qterm -- -p openai -b https://your-custom-endpoint.com/v1'
# qterm with MCP configuration
alias qtm='npm start --prefix /path/to/qterm -- -c /path/to/mcp-config.json'Add these lines to your shell configuration file:
# Basic qterm alias with Docker
alias qtd='docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm'
# qterm with specific provider and model
alias qtdg='docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -p google -m gemini-2.0-flash'
alias qtdo='docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -p openai -m gpt-4o'
# qterm with custom base URL
alias qtdc='docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -p openai -b https://your-custom-endpoint.com/v1'
# qterm with MCP configuration
alias qtdm='docker run -i --rm -v $(pwd)/.key:/app/.key:ro -v $(pwd)/mcp-config.json:/app/mcp-config.json:ro varlabz/qterm -c /app/mcp-config.json'After adding the aliases, reload your shell configuration:
# For Bash
source ~/.bashrc # or source ~/.bash_profile
# For Zsh
source ~/.zshrcOnce set up, you can use qterm with simple commands:
# Start interactive mode (npm version)
qt
# Start interactive mode (Docker version)
qtd
# Use with input (npm version)
qt -i "Your query here"
# Read input from stdin (npm version)
echo "Your query here" | qt -i -
# Use with input (Docker version)
qtd -i "Your query here"
# Read input from stdin (Docker version)
echo "Your query here" | qtd -i -
# Use Google's Gemini model (Docker version)
qtdg -i "Tell me about quantum computing"
# Use OpenAI with custom base URL (npm version)
qtc -i "What's the weather like today?"
# Use with MCP configuration (npm version)
qtm -i "Use the MCP tools to analyze this data"
# Use OpenAI with custom base URL (Docker version)
qtdc -i "What's the weather like today?"
# Use with MCP configuration (Docker version)
qtdm -i "Use the MCP tools to analyze this data"
# Use with access to local files (Docker version)
qtdl -i "Analyze the file in /workspace/myfile.txt"qterm integrates with two popular prompt libraries to provide a wide range of pre-built prompts for various use cases:
Fabric is a collection of prompts designed to solve specific problems using AI. Created by Daniel Miessler, Fabric provides a structured approach to prompt engineering with prompts organized by use case.
Fabric prompts are stored as Markdown files in the fabric/ directory and can be accessed using the -P option.
Awesome ChatGPT Prompts is a curated list of prompt examples for ChatGPT and other language models. It contains a wide variety of prompts for different roles and tasks.
Awesome ChatGPT Prompts are stored as text files in the awesome/ directory after being processed from the original CSV format.
You can use prompts from these libraries in several ways:
To see all available prompts from both libraries:
# Using npm
npm start -- -L
# Using Docker
docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -LThis will display a list of all prompts available in the fabric/, awesome/, and user/ directories.
To view the content of a specific prompt:
# Using npm
npm start -- -P "prompt-name"
# Using Docker
docker run -i --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -P "prompt-name"Replace prompt-name with the name of the prompt you want to view. You can use the name with or without the file extension.
To use a prompt from the libraries as a system prompt:
# Using npm
npm start -- -s "fabric/prompt-name.md"
# Using Docker
docker run -it --rm -v $(pwd)/.key:/app/.key:ro varlabz/qterm -s "awesome/prompt-name.txt"You can add your own custom prompts to the user/ directory. Create Markdown files with your prompts, and they will be automatically included when listing available prompts.
This project is licensed under the MIT License - see the LICENSE file for details.
The MIT License is a permissive license that allows for reuse with few restrictions. It permits use, modification, distribution, private use, and commercial use while providing no warranty or liability.