Code Style Prompt

prompt.txt

Reusable Code Style Prompt

When writing code for me, please follow these principles:

General Style Preferences

Prioritize clarity, correctness, and simplicity.

Code should be concise but not cryptic — use clear, descriptive variable names.

Avoid unnecessary complexity or over-engineering.

Maintain a logical structure with clear separation of concerns.


Formatting

Follow consistent formatting (e.g., PEP8 for Python, C# conventions for .NET).

Use spaces over tabs, standard indentation (2 or 4 spaces depending on the language).

Group code into well-commented sections, especially for multi-step logic.


Logging

Use structured and readable logs with emojis (where appropriate) to highlight key actions, successes, and errors.

Include:

Timestamps

Counts (rows, columns, bytes)

Time elapsed per major step

Summary stats at the end


For Python: use the logging module with a clean formatter, or structured logs via loguru or structlog if warranted.

For .NET: prefer ILogger and structured logging (e.g., Serilog or Microsoft.Extensions.Logging).


Libraries (Preferred)

Python: pandas, pyarrow, sqlalchemy, loguru, argparse, tqdm for progress.

C#: System.Text.Json, Serilog, HttpClientFactory, EF Core (when ORM needed).

Use standard libraries when possible; third-party only if necessary for clarity or performance.


Code Outputs and UX

If applicable, print friendly progress messages to console with:

Time elapsed (e.g., [01:23] ✅ Finished step 1)

Warnings/errors clearly marked (e.g., ⚠️ Warning: No records found)

Final summary report at the end.



Performance and Accuracy

Prioritize correctness first, but optimize for speed and memory where practical.

Use batching, streaming, or lazy evaluation for large data sets.

Ensure sorting, filtering, and calculations are explicit and deterministic.


Readability and Maintainability

Break code into functions or classes if logic exceeds 30–50 lines.

Use docstrings/comments to explain non-obvious logic, but don’t over-comment trivial steps.

Prefer returning structured results (e.g., dicts, namedtuples, or data classes) over printing raw output.