writing-style

Rewrites prose to remove AI writing patterns and match human expert style.

Updated Aug 5, 2026
One-click install
npx skills add https://github.com/harivansh-afk/loom-index-e2e --skill writing-style-harivansh-afk
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: writing-style
Source: https://github.com/harivansh-afk/loom-index-e2e/tree/main/skills/writing-style
Command: npx skills add https://github.com/harivansh-afk/loom-index-e2e --skill writing-style-harivansh-afk

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI-generated prose carries recognizable patterns (rule-of-three lists, em dashes, words like "delve" and "robust") that make readers stop trusting the text. This Skill gives writers and reviewers a concrete checklist and a reusable prompt to make docs, comments, issues, PRs, and chat replies read like a careful human expert wrote them. ## Core Features & Use Cases - AI-tell detection: Catalogs documented signs of AI writing across cadence, word choice, register, layout, and flow, each grounded in Wikipedia's Signs of AI writing field guide and cited vocabulary studies. - Rewriting rules: Provides concrete replacements, such as using "is" and "has" instead of "serves as", cutting summary endings, and breaking three-part symmetry. - Verbatim review prompt: Ships a copy-paste prompt with banned words and phrases for generating or auditing any draft. - Use Case: Before posting a PR description or README update, run the draft through the checklist to strip AI tells like negative parallelism, hedging filler, and decorative bullets. ## Quick Start Review my PR description draft and rewrite it to remove AI writing patterns using the writing-style rules.

Frequently Asked Questions about writing-style

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I make AI-generated text sound human?▼

Apply the checklist: cut rule-of-three structures, em dashes, and negative parallelism, replace banned words like delve and robust with plain verbs, and prefer "is" and "has" over "serves as". The Skill includes a verbatim prompt you can paste when generating or reviewing any draft.

What are the most common signs of AI writing?▼

The most recognizable tells are three-part parallel lists, "not X, but Y" constructions, spaced em dashes, rhetorical questions followed by answers, and summary endings. Word-level flags include delve, seamless, leverage, pivotal, and tapestry.

Which words should I avoid in technical writing?▼

Avoid the AI lexicon: delve, intricate, meticulous, pivotal, crucial, robust, seamless, leverage, foster, garner, showcase, underscore, testament, tapestry, vibrant, holistic, streamline, and elevate. One instance is a flag; two is a verdict.

Can I use this for code review comments and PR descriptions?▼

Yes. The Skill explicitly covers prose in docs, READMEs, comments, issues, PRs, and chat or review replies. For codebase answers it recommends giving the answer with a repo-relative path and quoting a snippet only when it is the clearest proof.

What sources back these AI writing detection rules?▼

The rules come from Wikipedia's Signs of AI writing field guide, Juzek and Ward's study on RLHF word overuse (arXiv:2508.01930), Kobak et al. on LLM vocabulary in academic abstracts, and Sam Kriss's NYT Magazine piece on AI writing style.