tux-gpt 0.2.4

An interactive terminal tool using GPT, with web search capabilities.

tux-gpt

tux-gpt is an interactive command-line assistant (not a background AI agent) that leverages GPT-based language models to provide intelligent, conversational help directly in your terminal. It enables on-the-fly code generation, debugging support, technical explanations, and more—all without leaving the command-line environment.

Designed for developers and tech enthusiasts, tux-gpt streamlines workflows by integrating AI assistance seamlessly into terminal sessions, making complex tasks easier and faster to accomplish via intuitive, context-aware command-line interactions. On startup the tool gathers details about the host operating system (distribution, architecture, etc.) so responses and suggested commands are tailored to your environment.

---

Prerequisites

• Python 3.7+
• Pip (Python package manager)
• An OpenAI API key (see next section)

---

Setup and Configuration

1. Install: From PyPI (for most platforms):

   pip install tux-gpt
   

   On newer Debian/Ubuntu releases where pip targets the system Python, prefer pipx to keep the install isolated:

   pipx install tux-gpt
   

   From source:

   git clone https://github.com/fberbert/tux-gpt.git
   cd tux-gpt
   pip install -r requirements.txt
   pip install .
   

2. Get your OpenAI API key:

   • Sign up or log in at https://platform.openai.com.
   • Navigate to API Keys and create a new key.
   • Copy the generated key.

3. Configure your environment variable:

   • Linux/macOS (bash/zsh):

     echo 'export OPENAI_API_KEY="<your_api_key>"' >> ~/.bashrc
     source ~/.bashrc
     

   • Windows (PowerShell):

     [Environment]::SetEnvironmentVariable('OPENAI_API_KEY', '<your_api_key>', 'User')
     

Note: On first run, tux-gpt will create the directory ~/.config/tux-gpt/ (or %APPDATA%\tux-gpt\ on Windows) containing:

• config.json: CLI configuration (e.g., default model);
• history.json: persistence of the last 20 messages (user + assistant);
• input_history: command history for navigation with ↑/↓ arrow keys.

---

Usage

Start the interactive session:

tux-gpt

Press Ctrl+J to send your message (Enter only inserts a new line).

Example commands

• Search the web for current news:

  > Find the latest headlines about OpenAI
  

• Look up technical documentation:

  > What is the syntax for Python's list comprehensions?
  

• Fetch real-time data (e.g., stock price):

  > What's the current stock price of AAPL?
  

• Summarize a web article:

  > Summarize the top result for "machine learning trends 2025"
  

• Run a single prompt from the shell:

  tux-gpt -q "What's the weather forecast for Rio das Ostras today?"
  

• Request JSON output for scripting:

  tux-gpt --json -q "Summarize the latest news about OpenAI"
  

  The response is a JSON object in the form:

  {
    "answer": "Short summary text...",
    "sources": ["https://example.com/article", "..."]
  }
  

• Generate and execute commands directly:

  tux-gpt -c "create a zip archive with every jpg or png in this folder"
  

  -c/--command asks Tux-GPT for a shell command. The assistant returns a JSON payload with the command plus a safety flag. If danger is false, the command runs immediately and the CLI prints Executando comando: .... If danger is true, the CLI shows the command and waits for confirmation before executing.

---

Memory & Command History

tux-gpt now persists your conversation and command history locally in the ~/.config/tux-gpt/ directory (or %APPDATA%\tux-gpt\ on Windows). The files created are:

• config.json: CLI configuration, such as the default model.
• history.json: stores the last 20 messages (user + assistant) to maintain context between sessions and limit token usage.
• input_history: command history used by readline for navigation with ↑/↓ arrow keys.

Features:

• On startup, the conversation history is automatically reloaded from history.json, limited to the last 20 messages to prevent token overload.
• You can navigate previous commands using the ↑ and ↓ arrow keys at the prompt.
• To reset the conversation or command history, simply remove the corresponding files in that directory (~/.config/tux-gpt/ or %APPDATA%\tux-gpt\).

---

Customization

You can configure the default model or terminal spinner settings by editing the configuration file at ~/.config/tux-gpt/config.json. Example:

{
  "model": "gpt-4o-mini"
}

---

Troubleshooting

• "OPENAI_API_KEY not set": Ensure you exported the variable correctly and restarted your shell.
• Slow responses: Check your internet connection or change to a faster model in the config.

---

License

MIT © 2025 tux-gpt contributors

---

Configuration File (~/.config/tux-gpt/config.json)

On the first run, tux-gpt will create a configuration file at ~/.config/tux-gpt/config.json (or %APPDATA%\tux-gpt\config.json on Windows). This file contains settings for the default model. You can customize the behavior of tux-gpt by editing the configuration file located at ~/.config/tux-gpt/config.json. This file allows you to set the default model and other preferences.

Example config file to set the model:

{
  "model": "gpt-5-mini"
}

The default model is gpt-5-mini.

Model Compatibility

Important: The model you choose must support web search capability. Currently, only the following models support the web search tool:

• gpt-5-mini
• gpt-4.1
• gpt-4.1-mini

For more details, see the official OpenAI documentation on web search tools and limitations:

https://platform.openai.com/docs/guides/tools-web-search?api-mode=responses#limitations

---

Author

Fábio Berbert de Paula

• Email: [e-mail]
• Website: https://fabio.automatizando.dev
• Founder of www.vivaolinux.com.br
• Over 25 years of experience as a developer

Official Repository

• https://github.com/fberbert/tux-gpt