Stop Slop is a skill file for removing AI-generated writing patterns from prose.
The tool addresses the problem that AI-written text exhibits recognizable patterns—predictable phrases, structural clichés, and rhythmic tells—that undermine authenticity and reader trust. It works by teaching Claude or other language models to identify and eliminate these patterns through a structured set of rules covering banned phrases like throat-clearing openers and business jargon, structural clichés such as binary contrasts and passive voice, and sentence-level constraints including restrictions on certain starters and punctuation choices. The skill includes a scoring system that rates prose across five dimensions: directness, rhythm, trust, authenticity, and density, with a combined score below 35 out of 50 indicating the need for revision.
The tool suits writers and editors who want to strip AI tells from their own work or from text they've received from others. It works with Claude through multiple integration paths—as a skill in Claude Code, as uploaded knowledge in Claude Projects, as custom instructions, or embedded in API system prompts. The approach is practical and granular, targeting specific phrases and structures rather than attempting broad stylistic transformation. Someone adopting this should understand it functions as a detection and editing guide rather than an automated rewriter; the scoring framework helps identify problem areas, but the actual revision work falls to the user.
The project shows substantial real-world adoption, with almost all open issues raised by outside users rather than the core team, indicating a meaningful base of people using it in practice and reporting genuine problems they encounter.
