/cleancode
/cleancode review # scan the whole project
/cleancode review <path> # scan a specific file or folder
/cleancode review --yes # skip the cost confirmation
/cleancode ignore # list ignored patterns
/cleancode ignore <pattern> # add an ignore glob
/cleancode paradigm # show or set the review paradigm
Scans the project for clean-code violations and writes a report to .aether/clean/report.md — point your AI assistant at it to apply the fixes. Aether does not rewrite your code itself.
How it works
Detection is hybrid, in two passes:
- Local heuristics scan every file for free — no API calls, no cost. They flag suspect files (long functions, deep nesting, magic numbers, and more).
- If an AI provider is configured, one batched call re-reviews only the files heuristics flagged, for a sharper pass. Clean files never touch the API.
Before that AI pass runs, Aether shows a cost/token estimate and asks for confirmation — the same flow as /genesis and /sync. Pass --yes to skip it.
Paradigms
/cleancode can review against different sets of principles:
| Paradigm | Focus |
|---|---|
clean-code (default) | Long functions, deep nesting, magic numbers, naming, duplication |
solid | Single responsibility, open/closed, dependency inversion, and the rest of SOLID |
functional | Immutability, pure functions, avoiding side effects |
google-style | Google's style guide conventions |
/cleancode paradigm solid
The chosen paradigm is stored per project in .aether/settings/cleancode-paradigm.json.
Ignoring paths
/cleancode ignore "**/*.gen.ts"
Ignore patterns are stored in .aether/settings/cleancode-ignore.json and are excluded from every future /cleancode review.
Output
Each run overwrites .aether/clean/report.md with the paradigm used, detection mode, and every issue found grouped by file — including a suggested fix for each one.
