Tala - Terminal AI Language Assistant

📥 Download Latest Release

Tala is a terminal-based AI language assistant built with Go. It provides a colorful, interactive interface for communicating with various AI providers including OpenAI, Anthropic, and Ollama, with intelligent file operations and shell command execution capabilities.

🎯 Available in multiple formats: DEB, RPM, Snap, AppImage, DMG, Windows executables, and standalone binaries for Linux, Windows, macOS, and FreeBSD.

Features

• Multiple AI Providers: Support for OpenAI, Anthropic Claude, and Ollama
• AI-Integrated File Operations: Natural language file and directory management
  • Create, read, update, delete files with simple commands
  • "Create test.txt with hello world" - and it actually creates the file!
• Secure Shell Command Execution: Run bash/shell commands safely
  • Built-in security filtering and timeout protection
  • Cross-platform support (Windows, Linux, macOS)
• Copy-Paste Friendly: No alt-screen mode - use your terminal's native copy/paste
• Colorful Interface: ANSI color support with semantic color coding for better readability
• Dual Interface Modes: Enhanced terminal (TUI) and graphical (GUI) interfaces
• Real-Time Stats: Live response times, token counts, and session statistics
• Concurrent Input: Type next commands while AI is processing responses
• Enhanced GUI: Dark theme, larger input fields, emoji integration, and progress indicators
• Local-First: Defaults to Ollama with deepseek-r1 (no API key required)
• Simple Configuration: JSON-based configuration with sensible defaults
• Terminal Native: Works with all standard terminal features and shortcuts

Quick Start

Prerequisites

• Go 1.24.4 or later
• For default setup: Ollama with deepseek-r1 model

Installation

git clone https://github.com/domykasas/tala
cd tala
go mod tidy

# Build TUI version (default)
go build -o tala

# Build GUI version
go build -tags gui -o tala-gui

First Run

# Install and start Ollama (if using default setup)
ollama serve
ollama pull deepseek-r1

# Run Terminal Interface (TUI)
./tala

# Run Graphical Interface (GUI)
./tala-gui

Configuration

Tala uses a JSON configuration file located at ~/.config/tala/config.json.

Default Configuration

{
  "api_key": "",
  "provider": "ollama",
  "model": "deepseek-r1",
  "temperature": 0.7,
  "max_tokens": 0,
  "system_prompt": "You are a helpful AI assistant."
}

Configuration Parameters Explained

• provider: AI service to use (ollama, openai, or anthropic)
• model: Specific AI model name for the chosen provider
• api_key: Authentication key (required for OpenAI/Anthropic, not needed for Ollama)
• temperature: Response creativity level (0.0-2.0)
  • 0.0: Very focused, deterministic responses
  • 0.7: Balanced creativity (recommended)
  • 2.0: Very creative, varied responses
• max_tokens: Maximum response length (0 = unlimited)
• system_prompt: Initial instruction for the AI assistant

Supported Providers

• Ollama (default): Local AI models, no API key required
  • Models: deepseek-r1, llama2, mistral, codellama, etc.
  • Setup: Install Ollama and pull desired model
• OpenAI: GPT models, requires API key
  • Models: gpt-3.5-turbo, gpt-4, gpt-4-turbo, etc.
  • Setup: Get API key from OpenAI platform
• Anthropic: Claude models, requires API key
  • Models: claude-3-sonnet, claude-3-haiku, claude-3-opus, etc.
  • Setup: Get API key from Anthropic console

Switching Providers

To use OpenAI or Anthropic, edit your config file:

{
  "api_key": "your-api-key-here",
  "provider": "openai",
  "model": "gpt-4",
  "temperature": 0.7,
  "max_tokens": 0,
  "system_prompt": "You are a helpful AI assistant."
}

Usage

Interface Controls

Terminal (TUI) Mode:

• Enter: Send message
• Ctrl+C: Quit application
• Ctrl+L: Clear screen and reset session stats
• Backspace: Delete characters from input

GUI Mode:

• Enter: New line in input field
• Shift+Enter: Send message
• Ctrl+N: New chat (clear history)
• Ctrl+Q: Quit application

File Operations

Tala understands natural language for file operations:

You: Create a file called notes.txt with "Hello World" in it
AI: ✓ Created file 'notes.txt'

You: List files in current directory  
AI: ✓ Listed directory contents

You: Run ls command
AI: ✓ Executed shell command successfully

Direct Commands

You can also use direct slash commands:

• /help - Show available commands and file operations
• /create <filename> <content> - Create files directly
• /ls - List directory contents
• /pwd - Show current directory

Copy and Paste

Tala is designed to work seamlessly with your terminal's copy-paste functionality:

• Select text: Use mouse or keyboard selection
• Copy: Use your terminal's copy shortcut (Ctrl+Shift+C, Cmd+C, etc.)
• Paste: Use your terminal's paste shortcut (Ctrl+Shift+V, Cmd+V, etc.)
• Scroll back: Use mouse wheel, PgUp/PgDn, or terminal scrollback

Statistics

Tala displays helpful statistics:

• Per-response: Token count and response time for each AI response
• Session-wide: Total requests, total tokens, and average response time
• Live loading: Real-time elapsed time while AI is thinking

Development

Running Tests

go test ./...

Project Structure

tala/
├── main.go              # Application entry point
├── config/              # Configuration management
│   ├── config.go        # Config struct and file operations
│   └── config_test.go   # Configuration tests
├── ai/                  # AI provider implementations
│   ├── provider.go      # Provider interface and implementations
│   ├── intent.go        # AI-powered intent detection
│   ├── tools.go         # Tool execution framework
│   └── *_test.go        # Comprehensive test suites
├── fileops/             # File system operations
│   ├── fileops.go       # CRUD operations for files/directories
│   ├── commands.go      # Direct command interface
│   └── *_test.go        # File operation tests
├── tui/                 # Terminal UI components
│   └── model.go         # Bubble Tea model implementation
├── go.mod              # Go module file
├── README.md           # This file
├── CHANGELOG.md        # Version history
├── ROADMAP.md          # Development roadmap
└── CLAUDE.md           # Development guide

Troubleshooting

Common Issues

1. Ollama connection failed: Ensure Ollama is running (ollama serve)
2. Model not found: Pull the model first (ollama pull deepseek-r1)
3. API key errors: Check your API key in the config file
4. Permission errors: Ensure config directory is writable

Debug Information

Set environment variable for verbose output:

export DEBUG=1
./tala

Contributing

We welcome contributions! Please see our development guide for details on:

• Code style and conventions
• Testing requirements
• Pull request process
• Development setup

License

MIT License - see LICENSE file for details.

Acknowledgments

• Ollama for local AI model serving
• The Go community for excellent tooling and libraries
• Terminal color standards and ANSI escape sequences for cross-platform compatibility
• The Unix philosophy of simple, composable tools

---

Tala - Making AI conversations as natural as terminal commands.