Python based command line interface for prompted conversation using ChatGPT API
ChatGPT-CLI Tool is a Python-based command-line interface (CLI) tool that allows users to interact with OpenAI's ChatGPT API efficiently. This tool provides several powerful features, including automated system role selection, code identification, and the ability to save identified code snippets as separate scripts. Additionally, the tool can scan previous conversation history for context and includes basic chain of thought tracking in prompts. Whether you're looking for insightful conversations, code suggestions, or a simple chat interface, this CLI tool streamlines your interactions with the ChatGPT API.
- openai >= 1.59.6
- Automated System Role Selection: Automatically assign system roles for your ChatGPT interaction, optimizing the model's responses based on your desired use case
- Code Detection: The tool automatically identifies code snippets in the responses from the ChatGPT model and formats them properly.
- Save Code as Separate Scripts: Detected code snippets can be saved as separate script files in your working directory for future use or execution.
- Flexible Command-Line Interface: Simple, yet powerful, CLI commands allow easy interaction with the OpenAI ChatGPT API.
- Iterative Response Iterpretation: Collects multiple responses to each query for model reflection, and condenses the best components into a single, higher quality response
- Chain of Thought Tracking: Adds prompts that track reasoning and thought process, improving responses in scenarios requiring step-by-step reasoning.
First, ensure you have Python 3.10+ installed on your system. You can install the ChatGPT-CLI tool directly from the repository.
Clone the repository:
pip install openai>=1.59.6
git clone https://github.com/mjenior/cli_assistant.git
cd cli_assistantThat's it! the CLI assistant does not require it's own installation to start submitting improved queries.
Optionally: [alias.py] begins a series of prompts to add a a customized bash alias to you profile to access the assistant with the command , this command can be run from any relative path.
Before using the tool, a final helpful step is to also set up your API keys. Otherwise you'll need to provide to the app directly (described below).
Set the environment variable(s):
export OPENAI_API_KEY="your_openai_api_key"
# Also will parse <DEEPSEEK_API_KEY> if is the requested model. Can also be provided directly, identical to OpenAI key.
That's it! The assistant API is ready to use in the termal, below are some example uses and additional arguments that further tailor responses.
Once installed, you can invoke the ChatGPT CLI tool directly from the terminal.
Example:
gpt "Help me compose an agenda for a week-long trip to Tokyo."All arguments:
REQUIRED
prompt : str
User prompt text, request that is sent to ChatGPT
OPTIONAL
role : str
System role text, predefines system behaviours or type of expertise
Several built-in options are available, refer to README for details
Default is assistant
model : str
LLM to use in queiries.
Default is gpt-4o
chain_of_thought : bool
Include chain of thought enforcement in user prompt.
Default is False
refine : bool
Automatically improve user prompt to improve query specificity.
Default is False
iters : int
WANRING: More testing required for reliability
Number of responses to generate and parse for model reflection
Default is 1
seed : str or int
Set moded seed for more deterministic reponses
Converts strings into binary-like equivalent, constrained by max system bit size
Default is based on the pinnacle code from Freakazoid
dim : str
Dimensions for Dall-e image generation
Default is 1024x1024
qual : str
Image quality for Dall-e images
Default is standard
unit_testing : bool
rite comprehesive unit tests for any generated code.
Default is False
key : str
User-specific OpenAI API key.
Default looks for pre-set OPENAI_API_KEY environmental variable.
verbose : bool
Print all additional information to StdOut
Default is False
silent : bool
Silences all StdOut
Default is False
urgent : bool
Add urgency to the request [UNTESTED].
Default is False
The --role option allows you to specify a system role for ChatGPT, which will optimize its responses based on the role you choose. Any text that does not match one of the existing role shortcuts will be submitted as a new custom role. The default is an improved personal assistant.
Available role shortcuts:
- assist (default): Standard personal assistant with improved ability to help with tasks
- compbio: Expertise in bioinformatics and systems biology. Knowledgeable in commonly used computational biology platforms.
- cancer: Specializing in cancer biology and molecular mechanisms, providing rigorous, evidence-based insights into cancer pathways, genetic factors, and therapeutic strategies
- dev: Senior python full stack developer with emphases in correct syntax, documentation, and unit testing.
- invest: Experience in technology stock investment and wealth management. Provides analyses for new stocks to invest in.
- art: Creates an images described by the prompt, default style leans toward illustrations
- photo: Gnereates more photo-realistic images
- story: Retells plot of popular books and movies to appropriate for ~3 year olds with fun changes to characters.
- rewrite: Writing assistant to help with clarity and brevity
Example 1:
assistant.py --role compbio --prompt "Generate a Python script to align DNA sequences and analyze the data. Add code to generate at least 2 figures summarizing the results."Example 2:
assistant.py --role "You are a Sr. game developer." --prompt "Recreate the game Chip's Challenge in python."The CLI tool automatically detects code snippets within ChatGPT's responses and formats them properly.
Example:
assistant.py --code True --prompt "Show me a Python function to find the maximum element in a list."Output:
def find_max(lst):
return max(lst)This assistant will then automatically save the generated code into find_max.time_stamp.py in the current working directory. Set to [True] by default.
This feature helps guide the model's response by breaking down the steps in complex reasoning tasks. The --thought flag enables the tool to append "chain of thought" prompts to ensure more detailed responses. It is [True] by default and automatically added to the default assistant, combio, developer, and invest system role prompts. The chain of thought flag will prompt the model to provide a step-by-step explanation or breakdown of reasoning, which can be especially useful in educational or technical explanations. It also helps mitigate the occurence of hallucinations.
Example:
assistant.py --chain_of_thought True --prompt "Can you write out a list of directions to change a tire?" Attempts to improve the clarity, focus, and specificity of a prompt to align with the desired outcomes or objectives. It involves adjusting language, structure, and scope to ensure the prompt effectively guides responses and generates accurate, relevant, and actionable results. Results are automatically submitted as a new query to the requested LLM.
Example:
assistant.py --refine True --prompt "Can you write out a list of directions to change a tire?" Result:
Can you provide detailed, step-by-step instructions for changing a tire, emphasizing key safety precautions and necessary tools? You should include comprehensive details like how to safely park the car, the importance of using a wheel chock, and the correct way to position the jack. Also, expand on how to properly remove the lug nuts, replace the tire, and ensure everything is secure before driving again.
This feature helps to increase the creative ability of a model thorugh multiple distinct reponse generation followed by critical evaluation for the most optimal response. The --iterations flag accepts an integer value representing the number of separate reponse iterations the model will create for the given prompt. Increasing this value past the 1 will prompt the model to also provide a summary of it's evaluation including why the returned response was selected over others. Tip: Best results might be seen increasing this number relative to the complexity of the input prompt, but diminishing returns do seem to occur at a certain point. WANRING: More testing required for reliability, so use with caution.
Example:
assistant.py --iters 3 --prompt "Create a python script to download DNA sequence data from the SRA, preprocess the data for maximum comparibility, and idenify putative gene sequences." You are able to set specific parameters of the output image created by Dall-e. Flags for length (--dim_l) and width (--dim_w) dimenions in pixels, as well as definition quality (--qual) have been implemented.
Example:
assistant.py --dim_l 800 --dim_w 600 --qual high --prompt "Please create an image of a cell dissolving into code in the style of the impressionists." You are also able to instead provide the key directly to the assistant if it is not specified by your system. The default settings attempt to pull from system-wide environmental variables ().
Example:
assistant.py --key YOUR_API_KEY_HERE --prompt "How do you make pizza dough?"If you encounter any problems, please file an issue along with a detailed description.
We welcome contributions! If you'd like to contribute to this project, please follow these steps:
- Fork the repository.
- Create a new feature branch (
git checkout -b feature-name). - Commit your changes (
git commit -m 'Add new feature'). - Push the branch (
git push origin feature-name). - Create a pull request.
This project is licensed under the MIT License. See the LICENSE file for more details.