run-code-analysis

name	run-code-analysis
description	Run code quality analysis on YAMLGraph. Use when: running ruff, bandit, radon, vulture, pytest coverage, checking code quality, finding dead code, security scanning, or generating improvement recommendations. For autonomous analysis, use the code-analysis agent instead.
argument-hint	target path like 'yamlgraph' or specific tool like 'ruff'

Run Code Analysis

Run automated code quality tools and generate recommendations. Canonical source: reference/code-analysis.md.

Note: The code-analysis agent can run this autonomously. This skill covers manual usage.

Quick Start (Graph-Based)

yamlgraph graph run examples/demos/code-analysis/graph.yaml \
  --var path="yamlgraph" \
  --var package="yamlgraph"

Individual Tools

Ruff (Linting + Style)

ruff check yamlgraph/                    # Check
ruff check --fix yamlgraph/              # Auto-fix
ruff format yamlgraph/                   # Format

Pytest (Tests + Coverage)

# Fast parallel (skip slow)
pytest tests/unit/ -q --no-cov -m "not slow" -n auto

# With coverage
pytest tests/ --cov=yamlgraph --cov-report=html

Bandit (Security)

bandit -r yamlgraph/ -ll -q

Flags: -ll = medium+ severity, -q = quiet (issues only).

Radon (Complexity)

radon cc yamlgraph/ -a -nc              # Cyclomatic complexity
radon mi yamlgraph/                     # Maintainability index

Target: functions < complexity 10. Module size: < 400 lines, max 450.

Vulture (Dead Code)

vulture yamlgraph/ vulture_whitelist.py

The whitelist file suppresses intentional unused exports.

Additional Checks

# Find TODOs/FIXMEs
grep -rn "TODO\|FIXME" yamlgraph/

# Line counts
find yamlgraph/ -name "*.py" | xargs wc -l | sort -n | tail -20

# Import boundary check
lint-imports

Code Quality Standards

Metric	Target
Module size	< 400 lines (max 450)
Function complexity	< 10 (radon cc)
Test coverage	≥ 80%
Security issues	0 medium+ (bandit)
Dead code	0 unreported (vulture)

Graph Structure

The code-analysis graph uses an agent node with shell tools:

nodes:
  run_analysis:
    type: agent
    prompt: code-analysis/analyzer
    tools: [run_ruff, run_tests, run_bandit, run_radon, run_vulture]
    max_iterations: 12
    state_key: analysis_results

  generate_recommendations:
    type: llm
    prompt: code-analysis/recommend
    requires: [analysis_results]
    state_key: recommendations

Prerequisites

pip install ruff bandit radon vulture