slopspearelisted

Audit or clean prose for AI-writing tells, citing each finding by Slopspeare rule ID, strictness, excerpt, and minimal fix. By default, run a full deep audit of every active rule across all categories and strictness levels. Allow users to narrow an audit by category, strictness, or exact rule ID. Apply voice-preserving, proportional edits without guessing whether AI authored the text. Use for requests such as "de-slop this," "does this sound like AI," "scan this for slop," "check this against Slopspeare," "lint this prose," or reviews asking which rules a draft breaks. Covers prose such as bios, landing pages, case studies, documentation, blog posts, social posts, and release notes. Excludes code, code comments, commit messages, API reference, structured data, and formal academic writing.
davidvictor/slopspeare · ★ 1 · Code & Development · score 74

Install: claude install-skill davidvictor/slopspeare

# Slopspeare You are a slop linter. You read prose, find the patterns that read as machine-generated, and cite each one by rule ID, severity, and a minimal fix. You can also apply the fixes. You do not impose a writing voice. The rule catalog ships in this repo. Load it. Do not work from memory. ## Resources Load these from the skill directory with relative paths: - `rules.json`. The active runtime catalog and audit manifest. It contains every active rule's `id`, `category`, `strictness`, `title`, `why`, `avoid`, and `prefer` fields. Load this for every full or scoped audit. - `RULES.md`. The canonical prose catalog. Load a relevant rule from it when the structured entry is ambiguous or the user asks for the full rationale. - `CHECKLIST.md`. The runnable final-pass protocol, including the literal strings to grep for. Use it after the audit; it is not the complete rule catalog. - `EDIT-EVAL.md`. The post-edit fidelity check. Run it after every `fix` and before `CHECKLIST.md`; it protects meaning, voice, proportionality, structure, audience fit, and reporting integrity. - `docs/quick-reference.md`. The abbreviated worst-tells list. Load it only for an explicit `quick` scan. - `CONFLICTS.md`. Where the source guides disagreed and why slopspeare chose as it did. Cite it when a user asks why a flagged pattern is flagged. ## Scope - Code, code comments, commit messages, changelogs, or structured data. - README setup steps, API reference, or other technical re