Sloptrim

Sloptrim is a command-line tool and agent plugin open-sourced by GitHub user seyedehsanhadi for spotting common AI-writing patterns. It scores a draft from 0 to 100 against 71 documented patterns; 62 have automatic detectors, and the rest are judged by the AI during the rewrite. In Claude Code it checks every draft saved through the editing tools, and when the score passes a threshold it hands the flagged sentences to the AI to rewrite. Other AI tools can use it through the init command, which writes the rules into a project's AGENTS.md or Cursor rules. It reads 20 formats, including Word, PowerPoint, Excel, EPUB and LaTeX, and the detector uses only Python's standard library, with no network and no model, so your text stays on your machine. What sets it apart is how plainly the author states its limits: the README and ethics notes both say the score describes the writing, cannot prove whether a model wrote something, and must not be used to judge an author.

Who it's for:people who write documents, articles or reports with Claude Code, Codex or Cursor and want the AI to strip out stock phrasing before handing over a draft.

Before you start: the detector needs Python 3.9 or later, and the automatic-check hooks need Node. Install the plugin in Claude Code, restart, and run /sloptrim doctor to check the install. Installing writes hook settings into your host, and the init command also edits the project's AGENTS.md and Cursor rule files. The README's published test data is English text, and it doesn't say how well it works on Chinese. PDF and RTF files aren't scored, and text inside images isn't read. The author notes there is no website or hosted version, and any site offering a service under this name is unrelated. The license is Apache-2.0.

Based on the project's public documentation. Not tested by AI Inspiration World.