ai-slop-cleaner
Regression-safe, deletion-first workflow that strips AI-generated noise from changed files.
AI-generated code often carries a layer of noise: redundant comments, over-explained variable names, unnecessary abstractions, and verbose scaffolding that was never needed. The ai-slop-cleaner skill runs a structured reviewer pass on changed files, deletes everything that does not earn its place, and re-runs verification to confirm nothing broke.
Purpose
ai-slop-cleaner enforces a deletion-first discipline on AI-authored output. Rather than polishing slop, it removes it. The skill targets recently changed files, applies a quality-reviewer pass with an explicit anti-slop checklist, makes edits, and validates with a regression run before declaring the result clean.
How to invoke
Natural language triggers: "deslop", "anti-slop", "clean up the AI noise", "remove slop".
Explicit slash: $ai-slop-cleaner
codex
> deslop the last set of changescodex
> anti-slop pass on src/api/What it does
- Identifies recently changed or staged files as the target set
- Applies a quality-reviewer checklist focused on redundancy, over-abstraction, and unnecessary verbosity
- Deletes or rewrites every line that does not serve a clear purpose
- Re-runs the project's test and lint suite to confirm regression safety
- Optionally runs in reviewer-only mode to report findings without making edits
Outputs
- Modified source files with slop removed
- Reviewer report listing each issue found and the action taken
- Fresh test and lint output confirming no regressions introduced
Related
$code-review— full code-reviewer pass for correctness and maintainability$review— composite router that includes quality and security passes$ralph— ralph runs a mandatory deslop pass before exiting