An AI agent skill that removes 20+ common AI-generated writing patterns from text while keeping the author's own voice.
conorbronsdon/avoid-ai-writing
An agent skill that audits and rewrites prose to remove common AI writing patterns, with detect-only and edit-in-place modes.
- Stars
- 5,094
- Forks
- 466
- Open issues
- 18
- Last commit
- 2 days ago
What is it?
avoid-ai-writing is a portable writing skill for coding and writing agents such as Claude Code, Codex, OpenClaw and Hermes. It flags patterns that make text read as machine-generated, including stock vocabulary, promotional wording, inflated significance, vague attributions, generic conclusions and list decoration, then rewrites the text with plainer alternatives.
The skill runs in three modes: rewrite, which returns one revised version; detect, which lists flags and separates real problems from judgment calls; and edit, which makes minimal changes directly in a prose file while refusing source code, configuration and generated data. An optional voice profile (casual, professional, technical, warm or blunt) sets the target tone.
A bundled zero-dependency detector runs in Node 18+ and the browser, and provides a CLI, a GitHub Action and a pre-commit hook that gate prose by finding count. The detector is calibrated for English; community adaptations in other languages are maintained separately.
Key features
- 01Rewrites text to replace AI-style vocabulary and phrasing with plainer wording.
- 02Detect mode lists flagged patterns and separates clear problems from judgment calls.
- 03Edit mode makes targeted changes to prose files in place and refuses code or config files.
- 04Optional voice profiles set the tone to casual, professional, technical, warm or blunt.
- 05Includes a zero-dependency detector with a CLI, GitHub Action and pre-commit hook.
Use cases
- Cleaning up blog posts, newsletters or documentation drafts before publishing.
- Auditing a team's written content for AI-sounding patterns without altering it.
- Gating prose quality in CI so that markdown files above a finding threshold fail the build.
- Reviewing GitHub issue or PR replies that read as overly formal or templated.
Quick start
- 1
Install the skill into your agent's skills directory by cloning the repository
git clone https://github.com/conorbronsdon/avoid-ai-writing ~/.claude/skills/avoid-ai-writing - 2
Invoke the skill in Claude Code with your text
/avoid-ai-writing <your text>
Things to know
- The rules and bundled detector are calibrated for English only.
Topics
FAQ
- What is avoid-ai-writing?
- An agent skill that audits and rewrites prose to remove common AI writing patterns, with detect-only and edit-in-place modes. avoid-ai-writing is a portable writing skill for coding and writing agents such as Claude Code, Codex, OpenClaw and Hermes. It flags patterns that make text read as machine-generated, including stock vocabulary, promotional wording, inflated significance, vague attributions, generic conclusions and list decoration, then rewrites the text with plainer alternatives.
- Is avoid-ai-writing free to use? What license does it have?
- avoid-ai-writing is open source under the MIT license. Check the license terms before using it in commercial products.
- Is avoid-ai-writing still maintained?
- Actively maintained. The last commit to avoid-ai-writing was on Oct 18, 2026.
- What can I use avoid-ai-writing for?
- Cleaning up blog posts, newsletters or documentation drafts before publishing. Auditing a team's written content for AI-sounding patterns without altering it. Gating prose quality in CI so that markdown files above a finding threshold fail the build. Reviewing GitHub issue or PR replies that read as overly formal or templated.
- How do I install avoid-ai-writing?
- Install the skill into your agent's skills directory by cloning the repository: git clone https://github.com/conorbronsdon/avoid-ai-writing ~/.claude/skills/avoid-ai-writing Invoke the skill in Claude Code with your text: /avoid-ai-writing <your text>
- Where can I get avoid-ai-writing?
- The source code and installation instructions are on GitHub: https://github.com/conorbronsdon/avoid-ai-writing
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
An Agent Skill that edits Chinese text to remove AI-style filler, repetition and templated phrasing while keeping meaning and facts.
A prompt-based skill that makes AI coding agents write only the code a task needs, favoring reuse and minimal solutions over large implementations.
A coding-assistant skill that makes AI agents lead with the next action, using numbered steps and no filler or preamble.
A portable Agent Skill that removes AI-writing tells from prose, repairing narrative structure in fiction and applying venue-specific rules to professional documents.
An AI-agent skill that generates editable slide decks in 12 visual themes, editable in the browser and exportable to HTML, PDF, and PPTX.