SamGalanakis

SamGalanakis / pilcrow

Public

A skill suite for writing

pilcrow.ink
32
0
89% credibility
Found May 19, 2026 at 35 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
TypeScript
AI Summary

Pilcrow is a writing quality checker that helps people polish their prose and remove AI-generated patterns from their text. It comes as a command-line tool you install once, then point at any document to get a detailed report of issues—everything from overused phrases and long sentences to the subtle patterns that make text sound robotic. The tool includes 49 automated checks plus 21 judgment-based rules, and can be extended with genre-specific rules for different types of writing like essays, emails, or documentation. You can use it just to find problems, or ask it to rewrite sections to sound more human and natural. It also integrates with AI coding assistants so you can check AI-generated documentation before publishing.

How It Works

1
📝 You want to check your writing

You've written something important—a blog post, documentation, or an essay—and you want to make sure it sounds natural and polished, not robotic.

2
🔧 You install the tool once

You run a simple install command, and the tool is ready on your computer. You can use it anytime without setting anything up again.

3
✨ You check your text

You point the tool at your writing, and it reads through everything—finding phrases that sound like AI wrote them, sentences that go on too long, and patterns that make prose feel stiff.

4
📋 You see a clear report

The tool shows you exactly what's flagged: line numbers, the problem phrases, and suggestions for fixing each one. Nothing is hidden or confusing.

5
You choose how to fix it
✍️
Fix it yourself

You use the suggestions as a guide and rewrite the flagged sections by hand, keeping your own voice intact.

🤖
Let it rewrite

You ask the tool to polish or humanize your text, and it rewrites the flagged parts while preserving what makes it yours.

🎉 Your writing is ready

The text sounds natural, flows well, and doesn't have the telltale signs of AI-generated prose. You're confident hitting publish or sending it out.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 35 to 32 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is pilcrow?

Pilcrow is a prose linter that catches AI-generated writing patterns and style guide violations in your markdown and text files. Built in TypeScript, it runs as both a standalone CLI tool and as an installed skill inside AI coding assistants like Claude Code, Cursor, and others. The engine combines 49 regex-based deterministic rules with 21 LLM-judged rules that evaluate semantic and rhetorical quality. You get human-readable audit output or JSON for programmatic consumption, plus genre-aware analysis across 39 document types from essays to error messages.

Why is it gaining traction?

The hook is dual: it catches the "AI tell" patterns that make LLM-generated prose read as robotic, and it enforces classical style guide discipline (Strunk, Williams, Zinsser, Pinker, King). Unlike generic grammar checkers, pilcrow targets the specific cadences that mark AI output: throat-clearing openers, antithesis rhythms like "It's not X, it's Y," hero-tagline imperatives, and corporate cliches. The skill integration is the real differentiator -- you can run `/pilcrow polish` directly inside your AI harness and get an ordered sequence of editor commands tuned to your document type. The voice capture system (`document`, `teach`, `craft` commands) lets it adapt findings to your actual writing style rather than applying generic rules.

Who should use this?

Technical writers who ship documentation and want to strip AI fingerprints from their output. Developers who use AI coding assistants and need to review the prose these tools generate. Content teams publishing READMEs, changelogs, or technical blog posts who want a second pass before shipping. Anyone whose writing now passes through an LLM and wants verification that it still sounds human.

Verdict

Pilcrow fills a real gap for teams dealing with AI-assisted writing, with a thoughtfully organized rule catalog and genuine utility in the skill integration. The credibility score of 0.8999999761581421% reflects a young project with only 32 stars -- it's early, the test coverage is unverified, and the genre taxonomy, while comprehensive, hasn't been battle-tested at scale. Install it globally, run it on your next AI-drafted README, and decide for yourself whether the output improves. At minimum, you'll learn which of your writing habits the model picked up.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.