find-complexity-hotspots

TODAY=!date +%Y-%m-%d

Complexity Hotspots

Identifies the most complex code in the codebase — by cyclomatic complexity, function length, and nesting depth — and produces targeted refactoring suggestions. Pairs naturally with /analyze-git-churn: files that are both frequently changed and structurally complex are the highest-priority refactoring targets.

Prerequisites

• Working directory is the root of the repository (or a subdirectory)
• Optional: $1 — path to limit the scan (defaults to .)
• Language-specific tools installed as available (skill degrades gracefully to heuristics if not present):
  • Python: radon (uv tool install radon or pip install radon)
  • JavaScript/TypeScript: complexity-report or eslint with complexity rule
  • Go: gocyclo (go install github.com/fzipp/gocyclo@latest)
  • Generic fallback: line-count and nesting-depth heuristics via rg/awk

Metrics

Metric	What it measures	Threshold to flag
Cyclomatic complexity	Number of independent paths through a function	≥ 10
Function length	Lines of code in a single function/method	≥ 50
Nesting depth	Maximum indentation depth inside a function	≥ 4 levels
File length	Total lines in a file	≥ 500

Flag any unit that exceeds any threshold. Rank by the number of thresholds exceeded, then by severity within each metric.

Steps

1. Detect Language and Available Tools

Identify the primary language(s) in the target path:

find ${1:-.} -type f | sed 's/.*\.//' | sort | uniq -c | sort -rn | head -10

Check for available analysis tools:

command -v radon && radon --version
command -v gocyclo && gocyclo -version

2. Measure Cyclomatic Complexity

Python — radon:

radon cc ${1:-.} -s -a -n C --total-average

Flag anything graded C (complexity 10–14), D (15–19), E (20–24), or F (25+).

Go — gocyclo:

gocyclo -over 10 ${1:-.}

JavaScript/TypeScript — eslint:

npx eslint ${1:-.} --rule '{"complexity": ["warn", 10]}' --format compact 2>/dev/null | rg complexity