A Claude Code skill that strips AI writing patterns from prose. Based on hardikpandya/stop-slop (MIT).
Scores text on 5 dimensions (directness, rhythm, trust, authenticity, density), identifies violations of 8 core rules, rewrites with all violations fixed, and re-scores.
Copy SKILL.md into a skill directory in your Claude Code project:
mkdir -p .claude/commands
cp SKILL.md .claude/commands/stop-slop.md
Or as a standalone skill directory:
mkdir -p skills/stop-slop
cp SKILL.md feedback.log skills/stop-slop/
/stop-slop Here's the email I need cleaned up: ...
/stop-slop rewrite this paragraph to sound human: ...
/stop-slop score this draft and fix anything that reads like AI
- Cut filler phrases
- Break formulaic structures
- Use active voice
- Be specific
- Put the reader in the room
- Vary rhythm
- Trust readers
- Cut quotables
Rated 1-10 on each dimension. Below 35/50 triggers a rewrite.
| Dimension | Question |
|---|---|
| Directness | Statements or announcements? |
| Rhythm | Varied or metronomic? |
| Trust | Respects reader intelligence? |
| Authenticity | Sounds human? |
| Density | Anything cuttable? |
The skill reads feedback.log before every execution and applies corrections from prior sessions. New corrections get appended during use. The skill improves over time.
MIT