CodepitStrip the AI tells out of your writing
Claude checks a draft for patterns common in AI-written text, like chatbot openers, inflated claims and filler, then rewrites it once in plainer words while keeping the facts.
Scan in CodepitScan checks all of conorbronsdon/avoid-ai-writing on your computer, before anything installs.
Codepit runs on your computer. Open this page there to scan it before installing.
Conor BronsdonMade by conorbronsdon
5.4kStars on GitHubStars
MITLicense
2 NovLast update
See it running
A real clip from the author's repo.
What you get
Three modes, plus optional voice profiles: casual, professional, technical, warm or blunt.
- Rewrite
- Detect only
- Edit a file in place
When it kicks in
- Remove AI-isms
- Audit for AI tells
- Make it sound less like AI
Say 'detect' or 'flag only' to get the findings, grouped by severity, without a rewrite.
Before you start
Nothing else to set up: the skill needs no external tools or API keys. The optional detector script needs Node 18 or later.
Try saying
Claude flags the patterns, edits in up to two passes and returns one final version, with a short list of changes and what it checked.
Worth knowing
The author calls it a writing-quality tool, not a verdict. The same patterns show up in human writing, especially from second-language writers, so don't use its flags alone to judge who wrote something.
Edit mode changes the file you name directly, with small targeted edits. It refuses source code, configuration and generated data.
