chatgpt

Start in chat mode

$ chatgpt

$ chatgpt [[-p|--prompt]] "[What is the regex to match an email address?]"

$ chatgpt [[-m|--model]] [gpt-4]

$ chatgpt [[-i|--init-prompt]] "[You are Rick, from Rick and Morty. Respond to questions using his mannerism and include insulting jokes.]"

$ echo "[How to view running processes on Ubuntu?]" | chatgpt

$ chatgpt [[-p|--prompt]] "[image: A white cat]"

chatgpt [options...] [prompt]

-h, --help
    Show help message and exit

-v, --version
    Display program version

-m, --model <MODEL>
    Select model, e.g., gpt-3.5-turbo or gpt-4 (default: gpt-3.5-turbo)

-r, --role <ROLE>
    Reply in specified role (e.g., coder, teacher)

-s, --stream
    Stream response in real-time

-p, --proxy <PROXY>
    Use specified proxy URL

-c, --clear
    Clear terminal screen before output

-M, --markdown
    Format output as Markdown

-o, --output <FILE>
    Save conversation to file (default: Untitled.md)

-t, --test
    Test API connection

-k, --key <KEY>
    Provide OpenAI API key directly

The chatgpt command provides an unofficial command-line interface to OpenAI's ChatGPT models, enabling seamless interaction from the Linux terminal. Ideal for developers, sysadmins, and power users, it supports querying AI for code assistance, explanations, or creative tasks without leaving the shell.

Installation typically involves Python's pip: pip install chatgpt-cli (from acheong08/ChatGPT repo). An OpenAI API key is required, obtainable from platform.openai.com. Basic invocation sends a prompt and streams or prints the response.

Key features include model selection (gpt-3.5-turbo or gpt-4), response streaming for real-time output, markdown formatting, conversation saving to files, proxy support for restricted networks, and role-playing modes. It maintains session context optionally via --continue, though primary use is stateless prompts.

Responses are colored and formatted for readability. Advanced scripting integrates it into pipelines, e.g., with jq for JSON parsing. While convenient, usage incurs API costs based on tokens processed.

Requires paid OpenAI API key; unofficial tool may break on API changes; respects rate limits and token costs; potential privacy risks with prompts.

pip install 'chatgpt[clip]' or clone from GitHub and pip install -e .; set OPENAI_API_KEY env var.

chatgpt 'Explain Linux pipes'
chatgpt -m gpt-4 -s -M 'Write a bash script'
chatgpt -t to verify setup.

Developed by acheong08 in late 2022; started as reverse-engineered web client, evolved to official OpenAI API support in v0.9+; widely forked with 10k+ GitHub stars.