Creating a High-Quality Python Development Prompt for GPT Agents

Learn how to craft an effective system prompt to produce high-quality Python code using GPT-4o and Claude 3.5 Sonnet LLM models.

Crafting a precise system prompt is crucial when using GPT agents for technical tasks, especially in Python development. I've found a specific system prompt that works well with GPT-4o and Claude 3.5 Sonnet LLM models, ensuring the generated code is clean, efficient, and production-ready.

Key Elements of the Prompt

Here's the prompt I use:

Act as an expert Python developer and help to design and create code blocks / modules as per the user specification.

- MUST provide clean, production-grade, high quality code.
- ASSUME the user is using python version 3.9+
- USE well-known python design patterns and object-oriented programming approaches
- MUST provide code blocks with proper google style docstrings
- MUST provide code blocks with input and return value type hinting.
- MUST use type hints
- PREFER to use F-string for formatting strings
- PREFER keeping functions Small: Each function should do one thing and do it well.
- USE @property: For getter and setter methods.
- USE List and Dictionary Comprehensions: They are more readable and efficient.
- USE generators for large datasets to save memory.
- USE logging: Replace print statements with logging for better control over output.
- MUST to implement robust error handling when calling external dependencies
- USE dataclasses for storing data
- USE pydantic version 1 for data validation and settings management.
- Ensure the code is presented in code blocks without comments and description.
- An Example use to be presented in if __name__ == "__main__":
- If code to be stored in multiple files, use #!filepath to signal that in the same code block.

Example Code

To illustrate the effectiveness of this prompt, here's an example of the Python code it generates.

Prompt was: Create an elegant function calculating fibonacci series


import logging
from typing import Generator, List
from pydantic import BaseModel, ValidationError


class FibonacciModel(BaseModel):
    value: int

def fibonacci_generator(limit: int) -> Generator[int, None, None]:
    a, b = 0, 1
    for _ in range(limit):
        yield a
        a, b = b, a + b

def validate_and_process_fibonacci(data: List[int]) -> List[int]:
    valid_data = []
    for item in data:
            model = FibonacciModel(value=item)
        except ValidationError as e:
            logging.error(f"Validation error for item {item}: {e}")
    return valid_data

def main(limit: int) -> List[int]:
    fibonacci_sequence = fibonacci_generator(limit)
    return validate_and_process_fibonacci(list(fibonacci_sequence))

if __name__ == "__main__":
    limit = 10
    result = main(limit)"Fibonacci series: {result}")


from typing import List
from functools import lru_cache

class FibonacciCalculator:
    def fibonacci(n: int) -> int:
        Calculate the nth Fibonacci number.

            n (int): The position of the Fibonacci number to calculate.

            int: The nth Fibonacci number.

            ValueError: If n is negative.
        if n < 0:
            raise ValueError("n must be a non-negative integer")
        if n <= 1:
            return n
        return FibonacciCalculator.fibonacci(n - 1) + FibonacciCalculator.fibonacci(n - 2)

    def generate_series(cls, length: int) -> List[int]:
        Generate a Fibonacci series of the specified length.

            length (int): The number of Fibonacci numbers to generate.

            List[int]: A list containing the Fibonacci series.

            ValueError: If length is negative.
        if length < 0:
            raise ValueError("Length must be a non-negative integer")
        return [cls.fibonacci(i) for i in range(length)]

if __name__ == "__main__":
    # Example usage
    series_length = 10
    fib_series = FibonacciCalculator.generate_series(series_length)
    print(f"Fibonacci series of length {series_length}: {fib_series}")

    # Calculate the 20th Fibonacci number
    n = 20
    fib_number = FibonacciCalculator.fibonacci(n)
    print(f"The {n}th Fibonacci number is: {fib_number}")