ChatGPT Terminal

A simple, fast, and practical ChatGPT wrapper you can use directly from your terminal.

This tool lets you talk to ChatGPT without opening a browser. This is great for quick explanations, coding help, and staying in flow.

---

Features

• Interactive REPL mode
• Prompts from the command line
• Input from stdin
• Persistent conversation memory
• Easy history reset
• No silent hangs

---

Usage

Interactive mode

-bash- chatgpt Example:

Welcome to the terminal of GPT (to quit -> 'exit')

explain bfs now compare it to dfs exit

One-shot prompt: -bash- chatgpt explain bfs

Pipe input: -bash- cat notes.txt | chatgpt

Reset conversation history: -bash- chatgpt --reset

You can also reset inside the REPL:

/reset

Installation:

1. Clone the repository -bash- git clone https://github.com/owenwolfe/ChatGPT-Terminal.git cd ChatGPT-Terminal

2. Create a venv -bash- python3 -m venv .venv source .venv/bin/activate

3. Install dependencies -bash- pip install openai

4. Set your OpenAI API key -bash- export OPENAI_API_KEY="sk-..." To make this permanent, add it to ~/.zshrc or ~/.bashrc.

5. Make the script executable -bash- chmod +x chatgpt.py

6. Install as a global command -bash- sudo ln -sf "$(pwd)/chatgpt.py" /usr/local/bin/chatgpt

Verify: -bash- which chatgpt Configuration Model selection By default, the wrapper uses: gpt-4.1-mini

You can override it: -bash- OPENAI_MODEL="gpt-4.1" chatgpt explain dijkstra

Conversation history history is stored at: -bash- ~/.chatgpt_history.json History is automatically trimmed to avoid excessive token usage.

If you see a 429 Too Many Requests error, it usually means: Your organization has hit its token-per-minute limit, or billing/credits are not configured

Sometimes you just want to type: -bash- chatgpt explain this error Without opening a browser or breaking focus.

This tool keeps ChatGPT where it’s most useful: your terminal.

License: MIT License

Acknowledgements: Built with help from ChatGPT (OpenAI).