pyneat-cli 2.2.1b0

LibCST-powered Python cleaner: security scanning, structural refactors, auto-fix, and PYNAGENT markers for AI editors (not just a formatter).

PyNEAT — Clean AI-Generated Python Code

The anti-spaghetti code cleaner — PyNEAT goes beyond black/ruff-style formatting: it runs AST + LibCST rules in a single pipeline to fix security issues, AI/legacy smell, and unsafe structure (nested if/else, bad except, mutable defaults, and more).

AI generates code fast. PyNEAT cleans what AI generates.

Install	One-liner
PyPI	pip install pyneat-cli
CLI	pyneat check file.py · pyneat clean file.py · pyneat clean-dir ./src

---

The Problem

AI coding assistants (Cursor, Copilot, Claude Code) generate code at incredible speed. But AI-generated code has predictable patterns that introduce security vulnerabilities and code quality issues:

• SQL injections in f-strings: f"SELECT * FROM users WHERE id = {user_id}"
• Hardcoded secrets: api_key = "sk_live_abc123..."
• Magic numbers: timeout = 300 # What does this mean?
• Empty except blocks: except: pass # Silent failures
• Resource leaks: open() without context manager

The Solution

PyNEAT = AI-Generated Code Preprocessor

AI generates code → PyNEAT cleans it → Clean code continues
       ↑                                    ↓
       └────────── PYNAGENT Markers ────────┘

Quick Demo

Before PyNEAT

# SQL Injection vulnerability
query = f"SELECT * FROM users WHERE id = {user_id}"
cursor.execute(query)

# Hardcoded secret
api_key = "sk_live_abc123xyz789secret"

# Magic number
timeout = 300  # What is 300?

# Empty except
try:
    risky_operation()
except:
    pass

After PyNEAT

# Parameterized query
query = "SELECT * FROM users WHERE id = ?"
cursor.execute(query, (user_id,))

# Environment variable
api_key = os.environ.get("API_KEY")

# Named constant
REQUEST_TIMEOUT_SECONDS = 300  # 5 minutes
timeout = REQUEST_TIMEOUT_SECONDS

# Proper error handling
try:
    risky_operation()
except Exception as e:
    logger.error(f"Operation failed: {e}")
    raise

Features

Feature	Description
50+ Security Rules	Catch SQL injection, command injection, secrets, weak crypto
AI Bug Patterns	Magic numbers, empty excepts, resource leaks, naming inconsistencies
LibCST structural surgery	Deep refactors (not whitespace-only): flatten arrow anti-pattern if/else, eval() handling, except: pass, mutable defaults, == None → is None, bad is on literals, type(x) == T → isinstance
Auto-fix	One command to clean entire codebase
Agent-to-Agent Handoff	PYNAGENT markers for AI editors (Cursor, Copilot, Claude)
LSP Integration	Real-time diagnostics in VS Code, Neovim, JetBrains
7-Layer Protection	Semantic guard, type shield, scope protection

Security Rules

Severity	Count	Examples
CRITICAL	9	SQL injection, command injection, RCE
HIGH	10	Hardcoded secrets, weak crypto
MEDIUM	20	SSRF, XSS, open redirect
LOW	10	Information disclosure
INFO	10	Best practice hints

Installation

pip install pyneat-cli

Optional: where a platform wheel ships the native scanner, installs can use the compiled extension for faster scans; otherwise the pure-Python + LibCST path still works.

pip install pyneat-cli

Quick Start

CLI

# Scan for security vulnerabilities
pyneat check file.py

# Auto-fix all issues
pyneat clean file.py

# Export manifest for AI editors
pyneat manifest file.py --format sarif

# Clean entire directory
pyneat clean-dir ./src

IDE Plugins

VS Code

1. Open VS Code → Extensions (Ctrl+Shift+X)
2. Search "PyNEAT"
3. Install

Neovim

-- init.lua
use 'pyneat/pyneat-vim'

JetBrains

1. Settings → Plugins → Marketplace
2. Search "PyNEAT"
3. Install

LSP Server

pip install pyneat-cli[server]
python -m pyneat.lsp

Agent-to-Agent Handoff

PyNEAT generates PYNAGENT markers that AI editors can read:

import os  # PYNAGENT: {"id":"PYN-001","type":"unused_import","severity":"medium","fix":"Remove"}

def main():
    pass

AI editors (Cursor, Copilot, Claude Code) read these markers and:

1. Understand what issues exist
2. Ask the user about intent
3. Fix code correctly

Architecture

┌─────────────────────────────────────────────────────────────┐
│                     IDE Layer                               │
│  VS Code  │  Neovim  │  Vim  │  JetBrains  │  CLI       │
└───────────┴──────────┴───────┴────────────┴───────────────┘
                           │
                           ▼
┌─────────────────────────────────────────────────────────────┐
│              PyNEAT Plugin Interface Layer                   │
│  LSP Server  │  CLI Commands  │  Manifest Export           │
└───────────────────────────┬───────────────────────────────┘
                            │
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                    PyNEAT Core Engine                       │
│  RuleEngine  │  AgentMarker  │  ManifestExporter           │
└─────────────────────────────────────────────────────────────┘

Ecosystem

Integration	Status
VS Code Extension	Complete
Neovim Plugin	Complete
JetBrains Plugin	Beta
LSP Server	Complete
GitHub Actions	Complete
GitLab CI	Complete
Pre-commit Hook	Complete

License

GNU Affero General Public License v3 (AGPLv3+)

Commercial licensing: [e-mail]