Just Prompt

🚀 just-prompt

just-prompt is a unified tool library dedicated to model prompts (prompts), supporting multiple large language models (LLMs), including OpenAI, Anthropic, Google Gemini, and DeepSeek. The library offers a rich set of functional modules, aiming to enhance users' control over models.

🚀 Quick Start

just-prompt is a versatile tool library designed to simplify the process of interacting with various large language models. It provides a unified interface for creating and managing prompts, enabling users to leverage the power of different models with ease.

✨ Features

• Multi-model Support: Compatible with popular large language models such as OpenAI (GPT-4, GPT-3.5-turbo), Anthropic Claude, Google Gemini, and DeepSeek.
• Context Injection: Allows embedding custom JSON context in prompts to enhance model understanding.
• Pluggable Components: Enables function expansion through the Atoms and Molecules mechanisms, supporting custom logic processing.
• Multi-language Support: Can detect the language of input text and automatically select the appropriate model for processing.

📦 Installation

pip install just-prompt

💻 Usage Examples

Basic Usage

from just_prompt import JustPrompt

# Initialize the tool and set API keys
jp = JustPrompt(
    openai_api_key="your_openai_key",
    anthropic_api_key="your_anthropic_key",
    google_api_key="your_google_key",
    deepseek_api_key="your_deepseek_key"
)

# Create a prompt
prompt = jp.Prompt(
    content="Write an article about artificial intelligence.",
    context={"user_id": "123", "preferences": ["technology", "applications"]}
)

# Get the response
response = prompt.get_response(model_name="openai:gpt-4")
print(response)

Advanced Usage

Context Injection

context = {
    "user_info": {"name": "张三", "age": 28},
    "query_params": {"temperature": 0.7, "max_tokens": 500}
}

prompt = jp.Prompt(
    content="Describe the user's hobbies based on the user information.",
    context=context
)

Pluggable Components

from just_prompt.molecules import ClaudeExtendedThinking

# Enable Claude's extended thinking function
extended_thinking = ClaudeExtendedThinking(tokens=1024)

prompt = jp.Prompt(
    content="Solve this math problem: 3 + 5 * 2",
    callbacks=[extended_thinking]
)

Multi-language Support

# Chinese prompt
prompt_zh = jp.Prompt(content="你好，你今天过得怎么样？")
response_zh = prompt_zh.get_response(model_name="deepseek")

# English prompt
prompt_en = jp.Prompt(content="Hello, how are you today?")
response_en = prompt_en.get_response(model_name="anthropic")

📚 Documentation

Project Structure

just-prompt/
├── just_prompt/         # Core tool library
│   ├── atoms/          # Atomic components: basic functional modules
│   │   ├── openai.py    # OpenAI model support
│   │   ├── anthropic.py # Anthropic model support
│   │   └── ...
│   ├── molecules/      # Molecular components: complex logic processing
│   │   ├── context_injector.py  # Context injection module
│   │   └── ...
│   └── __init__.py     # Tool library entry
├── examples/           # Usage examples
│   ├── basic_usage.py    # Basic usage example
│   ├── context_example.py  # Context injection example
│   └── extended_functionality.py  # Extended functionality example
└── README.md            # Project documentation

⚠️ Important Note

1. API Keys: Ensure that you correctly set the API keys for all models you use when initializing the tool.
2. Environment Dependencies: Some functions may require additional dependency packages, such as openai, anthropic, and google-generativeai.
3. Performance Optimization: Adjust prompt parameters (such as temperature and maximum tokens) according to your specific needs to achieve the best results.

For more detailed information, please refer to the project documentation and relevant resource links.