AI Slop Detection

AI Slop Detection

Slop is a density problem, not a word problem.

A single "delve" is fine. Five "delves" near a "tapestry" and an "embark" is generated text. This skill scores density per 100 words, marker clustering, and whether the overall register fits the document type. It does not ban words. It flags concentrations.

When NOT To Use

• Writing the replacement prose (use scribe:doc-generator)
• Drift against a specific voice profile (use scribe:voice-review)

Execution Workflow

Start by locating the findings, then read them.

uv run --with pyyaml python scripts/slop_score.py --audit <files>

One command over files or directories, one line per finding with a file and a line number, exit 0. It loads every category in data/languages/en.yaml, including the ones the merge gate declines to score: semicolon_splice, negative_definition, temporal_residue, contrastive_scaffold, over_explanation, the softer anthropomorphism verbs, and the per-document negation-density reading. Those print as (low) or (medium). Step 9's anti-goals apply to them: surface, never auto-apply.

A .py path is read as its comments and docstrings, with the code blanked and the line numbers kept. --python extends that to a directory sweep. The classification below already names code comments as a target; until the collector reached them, every category here was checked against markdown alone.

The steps below say what each category means and how to rewrite it. Run the script first so the steps have locations to work