Automated Prompt Engineering Hub

PromptBuilderAgent is an enterprise-focused, robust Python library designed for large-scale, multi-agent orchestration with LLMs (Large Language Models). It enables organizations to efficiently generate dynamic, task-specific prompts for LLM agents, with built-in features for validation, logging, and performance monitoring.

Key Features

• Multiple Prompt Generation: Create and execute prompts for multiple agents in parallel or sequential iterations.
• Enterprise-Grade Reliability: Integrated logging with Loguru for detailed tracking, performance monitoring, and error management.
• Pydantic-Based Validation: Ensure data integrity and type safety with powerful Pydantic validation.
• Scalable Execution: Run tasks across multiple iterations, capturing outputs, errors, and metadata for each run.
• JSON Prompt Storage: Save each task's prompt and output to structured JSON files for auditability and reproducibility.
• UUID and Timestamp Tracking: Every prompt is uniquely identified with a UUID and timestamp, providing full traceability for enterprise environments.
• Flexible Integration with LLM Agents: Seamlessly integrates with popular LLM models like GPT-4 for task execution, making it easy to deploy within any LLM-based architecture.

Installation

To get started, you can install the PromptBuilderAgent library via pip:

pip install prompt-builder-agent

You'll also need to set up the following environment variables:

1. OpenAI API Key: Set your OpenAI API key as an environment variable to allow the agent to communicate with GPT models.

export OPENAI_API_KEY="your-openai-api-key"

Alternatively, you can create a .env file and add the following:

OPENAI_API_KEY=your-openai-api-key

Quick Start

Here’s a quick example to help you get started with PromptBuilderAgent:

from loguru import logger
from ape.auto_prompter import PromptBuilderAgent

# Example usage:
if __name__ == "__main__":
    # Configure loguru to output to a file as well as the console
    logger.add("agent_log.log", rotation="10 MB")
    
    
    # Initialize the prompt builder agent
    agent = PromptBuilderAgent()

    # Run the task multiple times (iterations parameter)
    task = "Generate an system prompt for an agent analyzes balance sheets"
    agent_name = "Balance-Sheet-Agent-1"

    # Example: Run the task 3 times and save the results to JSON
    prompts = agent.run_task(
        task=task, agent_name=agent_name, iterations=3
    )

    # Display the prompts in the log
    for prompt in prompts:
        logger.info(
            f"Prompt: {prompt.prompt}\nOutput: {prompt.output}\n"
        )

Use Cases

PromptBuilderAgent is designed for enterprises and developers building multi-agent systems powered by LLMs. Key use cases include:

• Automated Prompt Generation: Streamline LLM-driven workflows by dynamically generating prompts based on tasks.
• Task-Specific Agents: Create agents tailored to specific business functions such as finance, marketing, and customer support.
• Auditability and Compliance: All prompts and outputs are saved to JSON with UUIDs and timestamps, providing a transparent audit trail.
• Scalability in AI Operations: Ideal for companies managing multiple agents or AI-driven processes that require repetitive tasks or workflows.

Advanced Features

Multiple Agent Support

PromptBuilderAgent allows you to generate prompts for multiple agents with ease. Each agent can have its own tasks, and you can run them concurrently or sequentially with iterations.

prompts = agent.run_task(task="Analyze stock performance", agent_name="Stock-Analysis-Agent", iterations=5)

Iterations for Task Execution

You can specify how many times a task should be executed, enabling A/B testing, prompt refinement, or batch task processing.

prompts = agent.run_task(task="Generate a financial report", agent_name="Finance-Agent", iterations=10)

JSON Storage

Each prompt and its corresponding output is saved as a JSON file with detailed metadata such as a UUID, timestamp, agent name, and more. This ensures full auditability of your LLM-driven operations.

{
    "id": "123e4567-e89b-12d3-a456-426614174000",
    "timestamp": "2024-09-24T12:34:56Z",
    "agent_name": "Finance-Agent",
    "task": "Generate a financial report",
    "prompt": "The user wants to generate a financial report...",
    "output": "Here is your financial report..."
}

Logging and Error Handling

PromptBuilderAgent leverages loguru for advanced logging and error management. All key operations, including prompt generation, agent execution, and errors, are logged for transparency and debugging.

Example log output:

2024-09-24 14:35:12 | INFO | Initialized PromptBuilderAgent successfully.
2024-09-24 14:35:13 | INFO | Running task 'Generate a financial report' for agent 'Finance-Agent'...
2024-09-24 14:35:15 | SUCCESS | Task executed successfully for iteration 1.

You can configure the logging to suit your enterprise needs by specifying different log levels, rotation policies, and formats.

Contributing

We welcome contributions from the open-source community and enterprise partners. If you're interested in improving this project or adding features, feel free to submit a pull request or raise an issue.

To get started:

1. Fork the repository.
2. Create your feature branch (git checkout -b feature/AmazingFeature).
3. Commit your changes (git commit -m 'Add some AmazingFeature').
4. Push to the branch (git push origin feature/AmazingFeature).
5. Open a pull request.

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Contact Us

For enterprise inquiries, custom integration support, or licensing, please contact:

Email: enterprise@promptbuilder.com
Website: promptbuilder.com

---

With PromptBuilderAgent, enterprises can unlock the full potential of LLM agents by automating prompt generation and scaling AI-driven workflows efficiently. Start using the tool today to supercharge your multi-agent system operations.

---