- Added offenders to
references/prose-phrases.md: elevate, shift, land, evolve/evolving, actionable, curious, and quietly (false understatement). - Added a
prose-marketing-verbsbenchmark fixture covering the new entries.
- Published the skill through the standard
skills/slop-cop/layout. - Added skills.sh installation commands for Codex, Claude Code, Cursor, and GitHub Copilot.
- Reframed the README around observable quality rather than authorship detection.
- Corrected the documented repository structure and invocation behavior.
- Added the zero-tolerance prose rule and refreshed public documentation.
- Moved the skill into
skills/slop-cop/so its parent directory matches its frontmatter name. - Narrowed trigger language and added a quality-not-authorship boundary.
- Reconciled punctuation, active voice, hedging, cadence, and scoring rules.
- Added unsupported-authority, fabricated-experience, invented-consensus, formatting-density, state, accessibility, dark-pattern, and code-security checks.
- Added benchmark fixtures, validation, hooks, CI, plugin metadata, security policy, contribution guide, linked credits, and reproducible packaging.