Diff-scoped AI-slop cleanup for sushiAI: strip narration comments, imagined-state guards, type laundering, one-use helpers and style drift from the current diff without changing behaviour. Use right before autoreview on any agent-written change.
Installs into .claude/skills of the current project.
Are you the author of Deslop?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/sushi-killer-deslop)
---
name: deslop
description: "Diff-scoped AI-slop cleanup for sushiAI: strip narration comments, imagined-state guards, type laundering, one-use helpers and style drift from the current diff without changing behaviour. Use right before autoreview on any agent-written change."
---
# Deslop
Clean only the current diff. Preserve behaviour absolutely.
1. Scope to `git diff $(git merge-base main HEAD)` plus untracked files. Never
run a repo-wide cleanup.
2. Inspect every changed hunk for:
- comments a maintainer would not write: narration of the edit, syntax
explanation, prose restating the code (keep comments that explain a
non-obvious constraint, ordering or platform quirk);
- `try`/`catch` or null checks guarding states that cannot occur, abnormal
for the surrounding module;
- casts that launder types (`as any`, `as unknown as T`);
- one-use variables or helpers that add no meaning;
- compatibility shims, aliases or fallbacks with no named contract;
- naming, imports or formatting that drift from the surrounding file,
raw hex colours where a token exists, Cyrillic in shipped source.
3. Fix a finding inline only when the cleanup is trivial and behaviour-neutral;
otherwise note it for the review.
4. Report in one to three sentences: what changed, what is left for review.
Run `$deslop` before `$autoreview`, never instead of it.
