Lint prose with regex and LLM rules across 30+ genres.
npx skills add https://github.com/SamGalanakis/pilcrow --skill pilcrow
Please help me install this Agent Skill. Skill: pilcrow Source: https://github.com/SamGalanakis/pilcrow/tree/main/.cursor/skills/pilcrow Command: npx skills add https://github.com/SamGalanakis/pilcrow --skill pilcrow
This Skill transforms a generic LLM into a specialized prose editor that catches AI tells, enforces classical style guides, and provides structured editing commands for markdown, HTML, and plain-text prose. It eliminates the gap between raw LLM output and polished, human-sounding writing by combining deterministic regex checks with LLM-judged style critiques across 30+ genres.
Use the pilcrow skill to review and polish your prose by running the pilcrow polish command on any markdown, HTML, or plain-text file to receive a triaged report with ship-blockers, worth-fixing items, and proposed rewrites anchored in classical style guides.
High-intent search queries and answers about installing and using this skill.
AI tell detection in generated prose works by combining 50 deterministic regex checks with 22 LLM-judged style rules. This hybrid approach catches weak verbs, nominalizations, and robotic phrasing, then proposes rewrites anchored in classical style guides.
Markdown linting can enforce style guides like Strunk & White, Williams, and Zinsser by applying specialized editor commands. These commands interpret prose findings through classical style lenses to produce triaged reports with proposed rewrites.
Voice profiling captures your unique writing style by building a voice profile from existing prose samples. The editor then uses this profile to ensure proposed rewrites match your personal signatures and tone across different documents.
Genre-aware prose editing adapts its rules and suggestions to 30+ genres including essays, postmortems, documentation, and marketing copy. Each genre applies specific demands, forbids, and tolerances to ensure appropriate style critiques.
Polishing and tightening LLM-generated HTML and plain-text involves running specialized editor commands like polish, tighten, and clarify. These commands apply deterministic and LLM-judged rules to produce triaged findings with ship-blockers and proposed rewrites.
Prose linting with deterministic regex rules is limited to pattern-based detection and cannot fully assess contextual style or tone. To address this, the editor supplements regex checks with 22 LLM-judged rules for deeper style critique and genre-specific analysis.
robertgussCapture writing voice from sample text into reusable YAML voice profiles.
z4nr34lDetect AI-writing patterns and rewrite text to sound natural.
igorferreiraAutomates Word document creation, editing, and analysis via OOPP.
Amazon-QuickApply a 13-point self-check to eliminate AI-generated writing patterns.
