I got tired of structural anti-patterns, so I built an AST-based code cleaner for Python 🧹

khanhnam-nathan — Sat, 04 Apr 2026 06:36:54 +0000

Hi DEV community! 👋

I spend a lot of time reviewing Python code. I noticed that while formatters (like Black or Ruff) are amazing at fixing styling and whitespaces, they don't fix structural logic flaws.

If you have a deeply nested if/else block (the Arrow Anti-pattern), or a dangerous eval(), standard formatters will just make that bad logic look pretty.

So, I decided to build PyNeat — a CLI tool based on Instagram's LibCST to perform deep structural refactoring.

🚀 What PyNeat Does
PyNeat scans your Python AST in a single pass and automatically refactors anti-patterns. Unlike the built-in ast module, because it uses LibCST, it preserves 100% of your original comments and whitespace.

Currently, the v1.0 MVP fixes:

The Arrow Anti-pattern: Flattens deeply nested if/else using guard clauses.

Dangerous Evals: Safely converts eval() to AST literals.

Empty Excepts: Detects silent except: pass failures.

Mutable Defaults: Fixes the infamous def func(items=[]) memory leak.

Identity Checks: Upgrades == None to is None.

💡 The Roadmap
It is still in its early MVP stage! I've already received some great feedback from Reddit about handling edge cases (like import hoisting), and I'm actively working on a robust pytest suite. The ultimate goal is a Rust rewrite (pyneat-rs) for massive multi-threading performance.

Links:

🐙 GitHub: https://github.com/khanhnam-nathan/Pyneat

📦 PyPI: pip install pyneat-cli

I would love to hear your thoughts! Have you ever struggled with writing structural refactoring tools? Any feedback on the architecture would be deeply appreciated!

DEV Community: khanhnam-nathan

I got tired of structural anti-patterns, so I built an AST-based code cleaner for Python 🧹