Pylint QA

SKILL: Python Code Quality Auditor

📝 Description

This skill allows the agent to perform static code analysis on Python files or directories using pylint. It identifies syntax errors, bugs, and stylistic departures from PEP 8, providing a numerical score (0.0 to 10.0) to quantify code health.

⚙️ Configuration

• Script Name: call_pylint.py
• Dependency: pylint
• Environment: Python 3.x

Prerequisites

Before execution, ensure the environment has the necessary package installed:

pip install pylint

🛠️ Implementation (call_pylint.py)

import subprocess
import sys
import re
import json

def lint_python_code(path: str):
    """
    Executes pylint on the provided path and returns a structured JSON result.
    """
    try:
        # Run pylint with exit-zero to prevent the shell from throwing an error code
        result = subprocess.run(
            [sys.executable, "-m", "pylint", path, "--exit-zero"],
            capture_output=True,
            text=True,
            check=True
        )
        
        output = result.stdout
        
        # Regex to capture the rating (e.g., 8.5/10)
        score_match = re.search(r"rated at (-?\d+\.\d+)/10", output)
        score = float(score_match.group(1)) if score_match else None
        
        response = {
            "success": True,
            "path": path,
            "score": score,
            "report": output.strip()
        }
        return json.dumps(response)

    except Exception as e:
        return json.dumps({"success": False, "error": str(e)})

if __name__ == "__main__":
    # Expects path as the first argument
    if len(sys.argv) > 1:
        print(lint_python_code(sys.argv[1]))

🚀 Usage Guide

Agent Instructions

The agent should invoke this skill whenever Python code is created, modified, or before a pull request is finalized.

Input Schema:

• path (string): The file path (e.g., "src/main.py") or directory (e.g., "tests/") to audit.

Output Schema:

• success (boolean): Whether the linting process finished.
• score (float): The code quality rating out of 10.
• report (string): The detailed list of violations and suggestions.

📋 Expected Behavior & Thresholds

[!TIP] If the score is low, the agent should look specifically for E (Error) and W (Warning) codes in the report first, as these impact functionality and safety more than stylistic conventions (C).