ChatX 1.0.0-alpha-20250421.1

ChatX - AI-powered CLI

ChatX is a command-line interface (CLI) application that provides a chat-based interaction with AI assistants. Built in C#, it leverages AI chat models from multiple providers with function calling capabilities to create a powerful tool for AI-assisted command-line operations.

Features

• Interactive AI Chat: Have conversations with an AI assistant directly in your terminal
• Multiple AI Providers: Support for OpenAI, Azure OpenAI, and GitHub Copilot APIs
• Provider Selection: Easily switch between different AI providers or use profiles
• Function Calling: Allow the AI assistant to execute various operations:
  • Run shell commands (Bash, CMD, PowerShell) with persistent sessions
  • Manipulate files (view, create, edit, replace text)
  • Access date and time information
• Comprehensive Configuration System:
  • Multiple configuration scopes (global, user, local)
  • Profile support for different configurations
  • Full configuration CLI with get, set, list, clear, add, remove commands
• Customizable Experience: Configure the AI's behavior with system prompts and other options
• Chat History: Load and save chat histories for later reference
• Command Aliases: Create shortcuts for frequently used command configurations
• Token Management: Automatically manages token usage for long conversations
• Chat Commands: Special commands like /clear, /save, /cost during chat sessions
  • MDX integration commands: /file, /files, /find, /search, /get, /run

Installation

Prerequisites

• .NET 8.0 SDK or later

Installing as a .NET Tool

Once published to NuGet, ChatX can be installed globally:

dotnet tool install --global ChatX --prerelease

Or locally in your current directory:

dotnet tool install --local ChatX --prerelease

After installation, you can run ChatX directly from your terminal:

chatx --input "Hello, how can you help me?"

Building from Source

1. Clone this repository:

   git clone https://github.com/robch/chatx.git
   cd chatx
   

2. Build the project:

   dotnet build
   

3. Run the application:

   dotnet run
   

Usage

Basic usage:

chatx [options]

Common Options

• --system-prompt <prompt>: Set a custom system prompt for the AI
• --input <text> or --question <text>: Provide input or questions to the AI
• --inputs <text...> or --questions <text...>: Provide multiple inputs to process sequentially
• --input-chat-history <file>: Load previous chat history from a file
• --output-chat-history <file>: Save chat history to a file
• --trim-token-target <n>: Set a target for trimming chat history when it gets too large
• --interactive: Control whether to enter interactive mode (default: true)
• --save-alias <name>: Save the current command options as a named alias
• --foreach var <name> in <values>: Define a loop variable with multiple values
• --foreach var <name> in <start>..<end>: Define a loop variable with a numeric range
• --help: Display help information

Examples

Start a chat session:

chatx

Ask a specific question:

chatx --input "How do I find the largest files in a directory using bash?"

Use a custom system prompt:

chatx --system-prompt "You are an expert Linux system administrator who gives concise answers."

Save and load chat history:

chatx --output-chat-history "linux-help-session.jsonl"
chatx --input-chat-history "linux-help-session.jsonl"

Environment Variables and Configuration

ChatX supports a flexible configuration system with multiple scopes (global, user, local) and formats (YAML, INI). You can:

1. Use environment variables
2. Use configuration files at:
   • Global: %ProgramData%\.chatx\config (Windows) or /etc/.chatx/config (Linux/macOS)
   • User: %AppData%\.chatx\config (Windows) or ~/.chatx/config (Linux/macOS)
   • Local: .chatx\config in the current directory

Common Configuration Settings

OpenAI API

• OPENAI_API_KEY: Your OpenAI API key
• OPENAI_CHAT_MODEL_NAME: Model name to use (default: gpt-4o)

Azure OpenAI API

• AZURE_OPENAI_API_KEY: Your Azure OpenAI API key
• AZURE_OPENAI_ENDPOINT: Your Azure OpenAI endpoint
• AZURE_OPENAI_CHAT_DEPLOYMENT: Your Azure OpenAI deployment name

GitHub Copilot API

• GITHUB_TOKEN: Your GitHub personal access token for Copilot API (preferred method)
• COPILOT_API_ENDPOINT: Copilot API endpoint (default: https://api.githubcopilot.com)
• COPILOT_MODEL_NAME: Model name to use (default: claude-3.7-sonnet)
• COPILOT_INTEGRATION_ID: Your Copilot integration ID (required for both auth methods)

Alternative HMAC authentication:

• COPILOT_HMAC_KEY: Your Copilot HMAC key

App Configuration

• CHATX_PREFERRED_PROVIDER: Set default AI provider (openai, azure-openai, copilot, copilot-hmac)

Configuration Profiles

You can create named profiles to store different provider configurations:

1. Create YAML files at .chatx/profiles/<name>.yaml
2. Use --profile <name> to load that profile

Example profile at .chatx/profiles/work.yaml:

app:
  preferredProvider: "azure-openai"

azure:
  openai:
    endpoint: "https://my-work-endpoint.openai.azure.com"
    chatDeployment: "gpt-4"

Configuration Management

Manage configurations from the command line:

chatx config list              # List all settings in current scope
chatx config list --any        # List all settings from all scopes
chatx config get KEY           # Get a configuration value
chatx config set KEY VALUE     # Set a configuration value
chatx config clear KEY         # Clear a configuration value

Documentation

For more detailed documentation, see:

• Getting Started
• Command Line Options
• Function Calling
• Creating Aliases
• Chat History
• Slash Commands - Special commands during chat sessions

License

Copyright(c) 2025, Rob Chambers. All rights reserved.

Chat Commands

ChatX supports several slash commands that can be used during interactive chat sessions:

Basic Commands

• /clear - Clear the current chat history
• /save - Save the current chat history to a file
• /cost - Show token usage and estimated cost of the session

MDX Integration Commands

• /file <pattern> - Search files matching pattern
• /files <pattern> - List files matching pattern
• /find <pattern> - Find occurrences of pattern in files
• /search <query> - Search the web for the given query
• /get <url> - Get and display content from a URL
• /run <command> - Run a command and display the result

These MDX integration commands require the MDX tool to be installed and available on your system.