skills.rest
jeremybrasher
jeremybrasherCOMMUNITY
Explore all skills by this author →
GitHub

avoid-ai-writing

Audit and rewrite prose to remove AI writing patterns across three modes.

Updated Aug 28, 2026
Search Tags:#editing#style#content cleanup#writing quality#text rewriting#ai writing detection#prose audit
One-click install
npx skills add https://github.com/jeremybrasher/grokbot-skills --skill avoid-ai-writing-jeremybrasher
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: avoid-ai-writing
Source: https://github.com/jeremybrasher/grokbot-skills/tree/main/skills/avoid-ai-writing
Command: npx skills add https://github.com/jeremybrasher/grokbot-skills --skill avoid-ai-writing-jeremybrasher

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI-generated text carries recognizable patterns — promotional language, filler phrases, uniform rhythm, chatbot artifacts — that make writing sound machine-produced. This Skill audits content for 62 pattern categories and a 112-entry tiered word replacement table, then rewrites it to read as human-authored prose. ## Core Features & Use Cases - Three operating modes: rewrite (default, with a second-pass audit), detect (flag-only with severity assessment), and edit (minimal in-place fixes to prose files, refusing code and config). - Tiered vocabulary detection: Tier 1A frequency markers always flag, Tier 1B clarity edits are weighted lower, Tier 2 flags in clusters, and Tier 3 flags only at high density. - Guardrails and HOLD conditions: refuses authorship verdicts, never invents facts or voice, and preserves quoted, fenced, tabulated, or attributed text. - Use Case: Paste a LinkedIn draft that reads as AI-generated; the Skill quotes each flagged pattern, returns a cleaned rewrite, summarizes the changes, and runs a second-pass audit to catch surviving tells. ## Quick Start Ask your agent to remove AI-isms from this draft and audit it for AI tells before rewriting.

Frequently Asked Questions about avoid-ai-writing

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

FAQPage Schema
How do I remove AI writing patterns from a draft?▼

Paste the text and ask the agent to remove AI-isms or clean up AI writing. The skill audits the draft against 62 pattern categories, quotes each flagged span, returns a rewritten version, and runs a second-pass audit on the result.

What is the difference between detect mode and rewrite mode?▼

Detect mode flags AI patterns without changing the text, separating clear problems from judgment calls. Rewrite mode additionally produces a cleaned version with a change summary and second-pass audit. Trigger detect mode with phrases like "flag only" or "scan."

Can this skill edit a file in place instead of returning rewritten text?▼

Yes, edit mode applies minimal targeted changes to a named prose file using the Edit tool, preserving already-human passages. It refuses source code, configuration, and generated data files because prose rewrites can corrupt structured content.

Does this skill work with Claude Code, Cursor, and OpenClaw?▼

Yes, the single SKILL.md follows the agentskills.io format and runs in Claude Code, Cursor, OpenClaw, Hermes, Codex, and other compatible agents. Install via npx skills add, git clone into the skills directory, or a plugin marketplace.

Can AI writing flags prove a text was machine-generated?▼

No. The skill explicitly treats patterns as signals, not proof, and refuses to produce authorship, hiring, or academic-integrity verdicts. Studies cited in the skill show high false-positive rates on non-native English writers, so flags must be paired with context.

Why does the skill refuse to add personality or voice to sterile text?▼

Its guardrails prohibit injecting fake first person, manufactured stakes, or invented specifics, since that installs a new detectable fingerprint. When a clean rewrite would be sterile, the skill returns a HOLD and asks the writer to supply the voice.

Related Skills

Explore More Content & Communication Skills→
udecodeudecode
1.2k

unslop

Draft, audit, and edit prose to remove AI-writing patterns while preserving facts and voice.

Official
Advanced
ahsanghalibahsanghalib

social-content

Drafts source-traceable social media posts, blogs, and approved visual assets for multiple channels.

Community
Intermediate
TravisLeeeeeeTravisLeeeeee
110

Proofreader

Proofreads grammar, spelling, punctuation, syntax, style consistency, and readability in documents.

Community
Intermediate
takusaotometakusaotome
5

ai-text-humanizer

Detect AI-like patterns in Japanese text and compute an AI-smell score.

Community
Intermediate