Skip to content

Latest commit

 

History

History
184 lines (135 loc) · 6.86 KB

README.md

File metadata and controls

184 lines (135 loc) · 6.86 KB
Raw

Terminal GPT (tgpt) 🚀

GitHub go.mod Go version GitHub release (latest by date) Arch Linux package Chocolatey Version

tgpt is a cross-platform command-line interface (CLI) tool that allows you to use AI chatbot in your Terminal without requiring API keys.

Currently available providers:

  • Blackbox AI (Blackbox model)
  • Duckduckgo (Supports several models)
  • Groq (Requires a free API Key. LLaMA2-70b & Mixtral-8x7b)
  • KoboldAI (koboldcpp/HF_SPACE_Tiefighter-13B)
  • Ollama (Supports many models)
  • OpenAI (All models, Requires API Key, supports custom endpoints)
  • Phind (Phind Model)

Image Generation Model: BlackBoxAi

Usage

Usage: tgpt [Flags] [Prompt]

Flags:
-s, --shell                                        Generate and Execute shell commands. (Experimental) 
-c, --code                                         Generate Code. (Experimental)
-q, --quiet                                        Gives response back without loading animation
-w, --whole                                        Gives response back as a whole text
-img, --image                                      Generate images from text
--provider                                         Set Provider. Detailed information has been provided below. (Env: AI_PROVIDER)

Some additional options can be set. However not all options are supported by all providers. Not supported options will just be ignored.
--model                                            Set Model
--key                                              Set API Key
--url                                              Set OpenAI API endpoint url
--temperature                                      Set temperature
--top_p                                            Set top_p
--max_length                                       Set max response length
--log                                              Set filepath to log conversation to (For interactive modes)
--preprompt                                        Set preprompt
-y                                                 Execute shell command without confirmation

Options:
-v, --version                                      Print version 
-h, --help                                         Print help message 
-i, --interactive                                  Start normal interactive mode 
-m, --multiline                                    Start multi-line interactive mode 
-cl, --changelog                                   See changelog of versions 
-u, --update                                       Update program 
--disable-input-limit                              Disables the checking of 4000 character input limit

Providers:
The default provider is phind. The AI_PROVIDER environment variable can be used to specify a different provider.
Available providers to use: blackboxai, duckduckgo, groq, koboldai, ollama, openai and phind

Provider: blackboxai
Uses BlackBox model. Great for developers

Provider: duckduckgo
Available models: gpt-4o-mini (default), meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo, mistralai/Mixtral-8x7B-Instruct-v0.1, claude-3-haiku-20240307

Provider: groq
Requires a free API Key. Supports LLaMA2-70b & Mixtral-8x7b

Provider: koboldai
Uses koboldcpp/HF_SPACE_Tiefighter-13B only, answers from novels

Provider: ollama
Needs to be run locally. Supports many models

Provider: openai
Needs API key to work and supports various models. Recognizes the OPENAI_API_KEY and OPENAI_MODEL environment variables. Supports custom urls with --url

Provider: phind
Uses Phind Model. Great for developers

Examples:
tgpt "What is internet?"
tgpt -m
tgpt -s "How to update my system?"
tgpt --provider duckduckgo "What is 1+1"
tgpt --provider openai --key "sk-xxxx" --model "gpt-3.5-turbo" "What is 1+1"
cat install.sh | tgpt "Explain the code"

demo

Installation ⏬

Download for GNU/Linux 🐧 or MacOS 🍎

The default download location is /usr/local/bin, but you can change it in the command to use a different location. However, make sure the location is added to your PATH environment variable for easy accessibility.

You can download it with the following command:

curl -sSL https://raw.githubusercontent.com/aandrew-me/tgpt/main/install | bash -s /usr/local/bin

If you are using Arch Linux, you can install with pacman:

pacman -S tgpt

FreeBSD 😈

To install the port:

cd /usr/ports/www/tgpt/ && make install clean

To install the package, run one of these commands:

pkg install www/tgpt
pkg install tgpt

Install with Go

You need to add the Go install directory to your system's shell path.

go install github.com/aandrew-me/tgpt/v2@latest

Windows 🪟

  • Scoop: Package installation with Scoop can be done using the following command:

    scoop install https://raw.githubusercontent.com/aandrew-me/tgpt/main/tgpt.json

  • Chocolatey

    choco install tgpt

Updating ⬆️

If you installed the program with the installation script, you may update it with

tgpt -u

It may require admin privileges.

Proxy

Support:

  1. environment variable

http_proxy or HTTP_PROXY with following available formats:

  • Http Proxy [ http://ip:port ]
  • Http Auth [ http://user:pass@ip:port ]
  • Socks5 Proxy [ socks5://ip:port ]
  • Socks5 Auth [ socks5://user:pass@ip:port ]
  1. configuration file

file location in the following order:

  • ./proxy.txt (in the same directory from where you are executing)
  • ~/.config/tgpt/proxy.txt

Example:

http://127.0.0.1:8080

From Release

You can download the executable for your operating system, rename it to tgpt (or any other desired name), and then execute it by typing ./tgpt while in that directory. Alternatively, you can add it to your PATH environmental variable and then execute it by simply typing tgpt.

Uninstalling

If you installed with the install script, you can execute the following command to remove the tgpt executable

sudo rm $(which tgpt)

Configuration file is usually located in ~/.config/tgpt on GNU/Linux Systems and in "Library/Application Support/tgpt" on MacOS