| name | deslop |
| description | Clean AI-generated slop from staged changes before commit |
Overview
Review staged changes (or specified scope) and remove common AI coding artifacts that reduce code quality. Reference guidelines/anti-patterns.md for the full list of patterns to identify.
Parameters
- scope (optional, default: "staged"): "staged", "diff", or "file:
"
Steps
1. Identify Slop Patterns
Scan the specified scope for these anti-patterns:
Constraints:
- You MUST flag excessive inline comments explaining obvious code
- You MUST flag paranoid try/catch blocks around internal code
- You MUST flag
Anytype casts that bypass type checking - You MUST flag backwards-compatibility hacks (renamed
_vars,# removedcomments) - You SHOULD flag placeholder TODOs without actionable context
- You SHOULD flag over-abstraction for single-use patterns
- You SHOULD flag unnecessary intermediate variables
- You MUST NOT remove comments that explain non-obvious business logic
- You MUST NOT remove error handling at system boundaries
2. Present Findings
List each instance with:
- File path and line number
- Category of slop (from anti-patterns.md)
- The problematic code snippet
- Suggested fix
Constraints:
- You MUST group findings by file
- You SHOULD prioritize MUST-fix items over SHOULD-fix items
3. Apply Fixes
Remove or simplify identified slop with user confirmation.
Constraints:
- You MUST preserve functional code behavior
- You MUST run
ruff formatafter changes - You MUST run
ruff checkto verify no new issues introduced - You SHOULD batch related fixes together
Examples
User: "/deslop" Agent: Reviews staged changes, identifies 3 excessive comments and 1 unnecessary try/catch, offers to clean them.
User: "/deslop file:src/auth.py"
Agent: Scans auth.py specifically, finds 2 Any casts and 1 over-engineered factory pattern, proposes simplifications.