ngpt 2.11.0

A lightweight Python CLI and library for interacting with OpenAI-compatible APIs, supporting both official and self-hosted LLM endpoints.

nGPT

A lightweight Python CLI and library for interacting with OpenAI-compatible APIs, supporting both official and self-hosted LLM endpoints.

Table of Contents

• Quick Start
• Features
• Documentation
• Installation
• Usage
  • Documentation
  • CLI Tool
  • Python Library
  • CLI Framework
• Configuration
  • Command Line Options
  • CLI Configuration
  • Interactive Configuration
  • Configuration File
  • Configuration Priority
• Contributing
• License

Quick Start

# Install with pip
pip install ngpt

# Or install with uv (faster)
uv pip install ngpt

# Or install globally as a CLI tool (recommended)
uv tool install ngpt

# Chat with default settings
ngpt "Tell me about quantum computing"

# Start an interactive chat session with conversation memory
ngpt -i

# Return response without streaming
ngpt -n "Tell me about quantum computing"

# Generate code
ngpt --code "function to calculate the Fibonacci sequence"

# Generate code with syntax highlighting
ngpt --code --prettify "function to calculate the Fibonacci sequence"

# Generate code with real-time syntax highlighting
ngpt --code --stream-prettify "function to calculate the Fibonacci sequence"

# Generate and execute shell commands
ngpt --shell "list all files in the current directory"

# Display markdown responses with beautiful formatting
ngpt --prettify "Explain markdown syntax with examples"

# Display markdown responses with real-time formatting
ngpt --stream-prettify "Explain markdown syntax with examples"

# Use a specific markdown renderer
ngpt --prettify --renderer=rich "Create a markdown table"

# Use multiline editor for complex prompts
ngpt --text

# Use custom system prompt
ngpt --preprompt "You are a Linux expert" "How do I find large files?"

# Log your conversation to a file
ngpt --interactive --log conversation.log

# Create a temporary log file automatically
ngpt --log "Tell me about quantum computing"

For more examples and detailed usage, visit the CLI Usage Guide.

Features

• ✅ Versatile: Use as a CLI tool, Python library, or CLI framework for building custom tools
• 🪶 Lightweight: Minimal dependencies with everything you need included
• 🔄 API Flexibility: Works with OpenAI, Ollama, Groq, and any compatible endpoint
• 💬 Interactive Chat: Continuous conversation with memory in modern UI
• 📊 Streaming Responses: Real-time output for better user experience
• 🔍 Web Search: Integrated with compatible API endpoints
• 🎨 Markdown Rendering: Beautiful formatting of markdown and code with syntax highlighting
• ⚡ Real-time Markdown: Stream responses with live updating syntax highlighting and formatting
• ⚙️ Multiple Configurations: Cross-platform config system supporting different profiles
• 💻 Shell Command Generation: OS-aware command execution
• 🧩 Clean Code Generation: Output code without markdown or explanations
• 📝 Rich Multiline Editor: Interactive multiline text input with syntax highlighting and intuitive controls
• 🎭 System Prompts: Customize model behavior with custom system prompts
• 📃 Conversation Logging: Save your conversations to text files for later reference
• 🧰 CLI Components: Reusable components for building custom AI-powered command-line tools
• 🔌 Modular Architecture: Well-structured codebase with clean separation of concerns

See the Feature Overview for more details.

Documentation

Comprehensive documentation, including API reference, usage guides, and examples, is available at:

https://nazdridoy.github.io/ngpt/

Key documentation sections:

• Installation Guide
• CLI Usage Guide
• Library Usage Guide
• CLI Framework Guide
• Configuration Guide
• Examples & Tutorials

Installation

# Installation with pip
pip install ngpt

# Or install with uv (faster installation)
uv pip install ngpt

# Or install globally as a CLI tool (recommended for command-line usage)
uv tool install ngpt

Requires Python 3.8 or newer.

For detailed installation instructions, see the Installation Guide.

Usage

As a CLI Tool

# Basic chat (default mode)
ngpt "Hello, how are you?"

# Interactive chat session with conversation history
ngpt -i

# Log conversation to a file
ngpt --interactive --log conversation.log

# Use custom system prompt to guide AI behavior
ngpt --preprompt "You are a Python programming tutor" "Explain decorators"

# Show version information
ngpt -v

# Show active configuration
ngpt --show-config

# Show all configurations
ngpt --show-config --all

# List available models for the active configuration
ngpt --list-models

# List models for a specific configuration
ngpt --list-models --config-index 1

# With custom options
ngpt --api-key your-key --base-url http://your-endpoint --model your-model "Hello"

# Enable web search (if your API endpoint supports it)
ngpt --web-search "What's the latest news about AI?"

# Generate and execute shell commands (using -s or --shell flag)
# OS-aware: generates appropriate commands for Windows, macOS, or Linux
ngpt -s "list all files in current directory"
# On Windows generates: dir
# On Linux/macOS generates: ls -la

# Generate clean code (using -c or --code flag)
# Returns only code without markdown formatting or explanations
ngpt -c "create a python function that calculates fibonacci numbers"

# Use multiline text editor for complex prompts (using -t or --text flag)
# Opens an interactive editor with syntax highlighting and intuitive controls
ngpt -t

For more CLI examples and detailed usage information, see the CLI Usage Guide.

As a Library

from ngpt import NGPTClient, load_config

# Load the first configuration (index 0) from config file
config = load_config(config_index=0)

# Initialize the client with config
client = NGPTClient(**config)

# Or initialize with custom parameters
client = NGPTClient(
    api_key="your-key",
    base_url="http://your-endpoint",
    provider="openai",
    model="o3-mini"
)

# Chat
response = client.chat("Hello, how are you?")

# Chat with web search (if your API endpoint supports it)
response = client.chat("What's the latest news about AI?", web_search=True)

# Generate shell command
command = client.generate_shell_command("list all files")

# Generate code
code = client.generate_code("create a python function that calculates fibonacci numbers")

For more library examples and advanced usage, see the Library Usage Guide.

Advanced Library Usage

# Stream responses
for chunk in client.chat("Write a poem about Python", stream=True):
    print(chunk, end="", flush=True)

# Customize system prompt
response = client.chat(
    "Explain quantum computing",
    system_prompt="You are a quantum physics professor. Explain complex concepts simply."
)

# OS-aware shell commands
# Automatically generates appropriate commands for the current OS
command = client.generate_shell_command("find large files")
import subprocess
result = subprocess.run(command, shell=True, capture_output=True, text=True)
print(result.stdout)

# Clean code generation
# Returns only code without markdown or explanations
code = client.generate_code("function that converts Celsius to Fahrenheit")
print(code)

For advanced usage patterns and integrations, check out the Advanced Examples.

As a CLI Framework

nGPT can also be used as a framework to build your own AI-powered command-line tools. You can leverage nGPT's pre-built CLI components to quickly develop sophisticated CLI applications.

from ngpt import NGPTClient, load_config
from ngpt.cli.interactive import interactive_chat_session
from ngpt.cli.renderers import prettify_markdown
from ngpt.cli.args import setup_argument_parser
import sys

# Create a custom CLI tool with colorized help
parser = setup_argument_parser()
parser.description = "Specialized Code Assistant"
parser.add_argument("prompt", nargs="?", help="Code description")
parser.add_argument("--language", "-l", default="python", help="Programming language")
parser.add_argument("--interactive", "-i", action="store_true", help="Start interactive mode")
args = parser.parse_args()

# Initialize client
client = NGPTClient(**load_config())

# Use interactive session for conversation
if args.interactive:
    system_prompt = f"You are an expert {args.language} developer. Provide clear, detailed answers."
    interactive_chat_session(client=client, preprompt=system_prompt, prettify=True)
elif args.prompt:
    # Generate and prettify code
    code = client.generate_code(args.prompt, language=args.language)
    print(prettify_markdown(f"```{args.language}\n{code}\n```"))
else:
    parser.print_help()
    sys.exit(1)

This allows you to build specialized AI tools like:

• Code generators for specific languages or frameworks
• Domain-specific assistants (SQL, legal, finance, etc.)
• Documentation generators
• Translation tools
• And much more

For detailed information about building CLI tools with nGPT, see the CLI Framework Guide and explore the CLI Component Examples.

Configuration

Command Line Options

You can configure the client using the following options:

Option	Description
--api-key	API key for the service
--base-url	Base URL for the API
--model	Model to use
--list-models	List all available models for the selected configuration (can be combined with --config-index)
--web-search	Enable web search capability
-n, --no-stream	Return the whole response without streaming
--temperature	Set temperature (controls randomness, default: 0.7)
--top_p	Set top_p (controls diversity, default: 1.0)
--max_tokens	Set maximum response length in tokens
--preprompt	Set custom system prompt to control AI behavior
--log	Enable logging: use --log to create a temporary log file, or --log PATH for a specific location
--prettify	Render markdown responses and code with syntax highlighting
--stream-prettify	Enable real-time markdown rendering with syntax highlighting while streaming
--renderer	Select which markdown renderer to use with --prettify (auto, rich, or glow)
--list-renderers	Show available markdown renderers for use with --prettify
--config	Path to a custom configuration file or, when used without a value, enters interactive configuration mode
--config-index	Index of the configuration to use (default: 0)
--provider	Provider name to identify the configuration to use (alternative to --config-index)
--remove	Remove the configuration at the specified index (requires --config and --config-index or --provider)
--show-config	Show configuration details and exit
--all	Used with --show-config to display all configurations
-i, --interactive	Start an interactive chat session with stylish UI, conversation history, and special commands
-s, --shell	Generate and execute shell commands
-c, --code	Generate clean code output
-t, --text	Open interactive multiline editor for complex prompts
--language	Programming language to generate code in (for code mode, default: python)
--cli-config	Manage CLI configuration settings (set, get, unset, list, help)
-v, --version	Show version information

For a complete reference of all available options, see the CLI Usage Guide.

CLI Configuration

NGPT offers a CLI configuration system that allows you to set default values for command-line options:

# Set default options
ngpt --cli-config set language typescript
ngpt --cli-config set temperature 0.9
ngpt --cli-config set prettify true

# View current settings
ngpt --cli-config get

# Get a specific setting
ngpt --cli-config get language

# Remove a setting
ngpt --cli-config unset prettify

# List all available options
ngpt --cli-config list

# Show help information
ngpt --cli-config help

Key features of CLI configuration:

• Context-Aware: Settings are applied based on the current command mode (e.g., language only applies in code generation mode -c).
• Priority: When determining option values, NGPT uses the following priority order (highest to lowest):
  1. Command-line arguments
  2. Environment variables
  3. CLI configuration (ngpt-cli.conf)
  4. Main configuration file (ngpt.conf)
  5. Default values
• Mutual Exclusivity: For options like no-stream, prettify, and stream-prettify, setting one to True automatically sets the others to False in the configuration file, ensuring consistency.
• Smart Selection: The provider setting is used to select which configuration profile to use, offering a persistent way to select your preferred API.

Available options include:

• General options (all modes): provider, temperature, top_p, max_tokens, preprompt, renderer, config-index, web-search
• Mode-specific options: language (code mode only), log (interactive and text modes)
• Mutually exclusive options: no-stream, prettify, stream-prettify

Practical Examples

# Set Gemini as your default provider
ngpt --cli-config set provider Gemini
# Now you can run commands without specifying --provider
ngpt "Explain quantum computing"

# Configure code generation for TypeScript
ngpt --cli-config set language typescript
# Now in code mode, TypeScript will be used by default
ngpt -c "Write a function to sort an array"

# Set a higher temperature for more creative responses
ngpt --cli-config set temperature 0.9

The CLI configuration is stored in:

• Linux: ~/.config/ngpt/ngpt-cli.conf
• macOS: ~/Library/Application Support/ngpt/ngpt-cli.conf
• Windows: %APPDATA%\ngpt\ngpt-cli.conf

For more details, see the CLI Configuration Guide.

Interactive Configuration

The --config option without arguments enters interactive configuration mode, allowing you to add or edit configurations:

# Add a new configuration
ngpt --config

# Edit an existing configuration at index 1
ngpt --config --config-index 1

# Edit an existing configuration by provider name
ngpt --config --provider Gemini

# Remove a configuration at index 2
ngpt --config --remove --config-index 2

# Remove a configuration by provider name
ngpt --config --remove --provider Gemini

# Use a specific configuration by provider name
ngpt --provider OpenAI "Tell me about quantum computing"

In interactive mode:

• When editing an existing configuration, press Enter to keep the current values
• When creating a new configuration, press Enter to use default values
• For security, your API key is not displayed when editing configurations
• When removing a configuration, you'll be asked to confirm before deletion

For more details on configuring nGPT, see the Configuration Guide.

Configuration File

nGPT uses a configuration file stored in the standard user config directory for your operating system:

• Linux: ~/.config/ngpt/ngpt.conf or $XDG_CONFIG_HOME/ngpt/ngpt.conf
• macOS: ~/Library/Application Support/ngpt/ngpt.conf
• Windows: %APPDATA%\ngpt\ngpt.conf

The configuration file uses a JSON list format, allowing you to store multiple configurations. You can select which configuration to use with the --config-index argument (or by default, index 0 is used).

Multiple Configurations Example (ngpt.conf)

[
  {
    "api_key": "your-openai-api-key-here",
    "base_url": "https://api.openai.com/v1/",
    "provider": "OpenAI",
    "model": "gpt-4o"
  },
  {
    "api_key": "your-groq-api-key-here",
    "base_url": "https://api.groq.com/openai/v1/",
    "provider": "Groq",
    "model": "llama3-70b-8192"
  },
  {
    "api_key": "your-ollama-key-if-needed",
    "base_url": "http://localhost:11434/v1/",
    "provider": "Ollama-Local",
    "model": "llama3"
  }
]

For details on the configuration file format and structure, see the Configuration Guide.

Configuration Priority

nGPT determines configuration values in the following order (highest priority first):

1. Command line arguments (--api-key, --base-url, --model, etc.)
2. Environment variables (OPENAI_API_KEY, OPENAI_BASE_URL, OPENAI_MODEL)
3. CLI configuration file (ngpt-cli.conf, managed with --cli-config)
4. Main configuration file ngpt.conf or custom-config-file
5. Default values

Contributing

We welcome contributions to nGPT! Whether it's bug fixes, feature additions, or documentation improvements, your help is appreciated.

To contribute:

1. Fork the repository
2. Create a feature branch: git checkout -b feature/your-feature-name
3. Make your changes
4. Commit with clear messages following conventional commit guidelines
5. Push to your fork and submit a pull request

Please check the CONTRIBUTING.md file for detailed guidelines on code style, pull request process, and development setup.

License

This project is licensed under the MIT License. See the LICENSE file for details.