ai-agents-metrics

ARCH-011: Integrate radon for code complexity metrics

Status: done
Priority: medium
Complexity: low

Rationale

Automated code quality metrics help identify overly complex functions and modules early. radon provides:

• Cyclomatic complexity — detects functions with too many decision paths (high complexity > 10 suggests SRP violation)
• Maintainability Index — aggregated score (0–100) combining LOC + complexity + Halstead metrics
• Lines of Code (LOC) — identifies modules that are too large and need decomposition

Current state: No automated complexity tooling beyond pylint and mypy.

Implementation

1. Add radon to dev dependencies in pyproject.toml
2. Add radon check to Makefile (under a new quality target or extended verify)
3. Configure radon threshold in pyproject.toml or .radon.ini:
   • Cyclomatic complexity warning threshold: 10 (flag “complex” functions)
   • Maintainability Index warning: < 40 (flag hard-to-maintain modules)
4. Integrate into CI/CD: fail on complexity violations or add to pre-push hook
5. Document in CLAUDE.md how to run locally: radon mi src/ -s

Acceptance Criteria

• radon is installed and runs without errors on src/
• make complexity — advisory report of all rank C+ functions across src/
• make complexity-check — global hard gate: fail if any function exceeds CC 50
• make verify includes both complexity and complexity-check

This site is open source. Improve this page.