Детектор шаблонного AI-текста

vibe-coding

slop-guard is a Claude Code skill that acts as a rule-based prose linter, scoring text from 0 to 100 for formulaic AI writing patterns with no LLM judge and no API calls — purely programmatic. The default pipeline runs 24 configurable rules backed by 200+ literal and structural heuristics, returning a numeric score, a band label, specific violations with surrounding context, and concrete advice for each hit. Connect it to Claude Code with `claude mcp add slop-guard -- uvx slop-guard`; the `sg` CLI lets you lint files, stdin, and inline text directly from the terminal. Detected patterns include stock hype phrases, assistant tone markers, unattributed weasel phrasing, bullet-heavy Markdown structures, em dash overuse, and uneven paragraph cadence. Developers and technical writers who need to gate prose quality in CI pipelines or editorial workflows can set a passing threshold with `--threshold` and fail builds automatically on low-scoring files.

• #qa