Code health

The code is formatted, clean and stays maintainable.

Each level adds measures; a level is reached when its measures and all measures below meet their thresholds. Measures and levels are fixed by the danwos model (The fixed danwos model deci... (DEC_DANWOS_MODEL)); a project only says where the results come from.

Level 1 – Sketched

Formatter configured, clean.

Level 2 – Shaped

Lint and type check clean.

Level 3 – Solid

Complexity and duplication within limits.

Level 4 – Refined

Static analysis 0 high findings, dependency rules hold.

Level 5 – Audited

Every change reviewed (human or second agent).

Possible checks

Examples – which check feeds which measure is the project’s choice.

• formatter check: prettier, ruff format

• lint and types: eslint + tsc, ruff + mypy

• complexity and duplication: e.g. lizard, jscpd

• static analysis: semgrep, CodeQL (SARIF)

• dependency rules: dependency-cruiser, import-linter

• review evidence per change