Ask GPT

English | 简体中文

Ask GPT is a CLI tool to chat with OpenAI LLMs via API.

Features

• Custom Roles: Customise multi roles, each with its own system prompt.
• Save/Retrieve Conversations: Save conversations locally and continue it seamlessly in the future.
• Streaming Output: See the output instantly after an input, no need to wait for completion.
• Support Pipeline: Send any output of other CLI tools to AI via pipeline.
• Interactive mode: Chat with AI interactively. Offer a user experience similar to the web-based ChatGPT.
• Support Models: Compatibility with all models accessible through the OpenAI API.
• Edit Config Easily: One-click access to configuration file for easy setup and customization.

Installation

Download the pre-built binary from release and add it to path.

Usage

1. Run ask-gpt for the first time:

   ag -v
   

   and the app will create the config file at

   %USERPROFILE%/.config/ask-gpt/config.yaml
   

   for Windows and

   $HOME/.config/ask-gpt/config.yaml
   

   for Linux and macOS.

2. Edit the config. See Configuration for detail.

3. Quickly ask a question:

   ag "what's the best programming language in the world?"
   

4. To create or continue a conversation, use -s:

   ag -s "chat 1" "Tell a story"
   

5. To chat interactively, use -i:

   ag -i "Translate the given sentences to Chinese"
   

   -s can be used with -i:

   ag -s "translation" -i "Translate the given sentences to Chinese"
   

6. Use pipeline:

   cat raw.txt | ag -i "Translate above contents to Chinese"
   

7. To create a new role, use -n:

   ag -n
   

   then set its name, description and prompt interactively.

8. To use a role, use -r (the role should already exist):

   ag -r "translator" -i "Translate the given sentences to Chinese"
   

   -r can be used with -s but only valid when creating a conversation.

9. To show history of the conversation, use -h:

   ag -s "translation" -h -i
   

   this will print previous messages and continue with interactive mode.

10. For more usage, see:

    chatgpt --help
    

Configuration

Name	Meaning	Default
url	The base URL for OpenAI API.	"https://api.openai.com/v1"
api_key	Your OpenAI API key.	""
model	The model to be used.	"gpt-4o-mini"
max_tokens	The maximum token cost in a single API call.	4096
temperature	See OpenAI Doc	0.5
top_p	See OpenAI Doc	1.0
frequency_penalty	See OpenAI Doc	0
presence_penalty	See OpenAI Doc	0
editor	The editor to be used to edit config file. Must in path.	"code"
editor_arg	The arguments to be passed to the editor. %path will be replaced by the actual path.	"%path"