thoo avatar

python-code-review

Review Python code changes in merge requests and pull requests

Install

/plugin install python-code-review@my-agent-skills

Run it inside Claude Code after adding the marketplace with /plugin marketplace add thoo/agent-skills.

Contents

Discovered from source

Skills 1

Python Code Review

Review Python code changes systematically using established best practices and checklists.

Review Workflow

1. Gather Context

# Get the diff to review
git diff main...HEAD

# Or for a specific PR
gh pr diff <pr-number>

2. Run Automated Checks First

Before manual review, verify automated tools pass:

ruff check .                    # Linting
ruff format --check .           # Formatting
mypy src/                       # Type checking
pytest tests/ -v                # Tests

3. Systematic Code Review

Review each file in the diff, checking these categories in order:

Security (Critical)

  • No SQL string formatting (use parameterized queries)
  • No eval()/exec() with user input
  • No hardcoded secrets
  • Inputs validated at boundaries
  • No path traversal vulnerabilities

Correctness

  • Logic handles edge cases (empty, None, boundaries)
  • Error handling is specific (no bare except:)
  • Exception chains preserved (raise ... from e)
  • No mutable default arguments
  • Types are correct and complete

Design

  • Functions do one thing (<30 lines)
  • Classes follow single responsibility
  • No code duplication (DRY)
  • Appropriate data structures used
  • No premature optimization

Style

  • Naming is clear and consistent (snake_case, PascalCase)
  • No magic numbers/strings
  • Docstrings on public API
  • No commented-out code

4. Provide Feedback

Structure feedback by severity:

  1. Blockers - Must fix before merge (security, bugs, breaking changes)
  2. Suggestions - Should consider (design, performance, readability)
  3. Nits - Minor style issues (optional to fix)

Red Flags to Catch

PatternIssue
except: or except Exception:Catches too much, hides bugs
def foo(items=[])Mutable default argument bug
f"SELECT * FROM {table}"SQL injection
eval(user_input)Code injection
password = "secret123"Hardcoded secret
No type hints on public functionsHarder to maintain
Functions >50 linesLikely doing too much
# TODO without ticketTech debt without tracking

Common Patterns to Suggest

# Instead of bare except
try:
    process()
except:  # Bad
    pass

# Use specific exceptions
try:
    process()
except ValueError as e:
    logger.exception("Processing failed")
    raise

# Instead of mutable defaults
def add_item(item, items=[]):  # Bad
    items.append(item)

# Use None pattern
def add_item(item, items: list | None = None) -> list:
    if items is None:
        items = []
    items.append(item)
    return items

# Instead of verbose loops
result = []
for x in data:
    if x.valid:
        result.append(x.value)

# Use comprehensions
result = [x.value for x in data if x.valid]

Reference Guides

For detailed guidelines, consult:

Use these references when:

  • Explaining why something should change
  • Providing examples of better patterns
  • Checking specific categories in depth

Sponsors

Details

Marketplace
thoo/agent-skills
Category
Community